Physiol Mol Biol Plants (February 2020) 26(2):317–330
https://doi.org/10.1007/s12298-019-00740-x
RESEARCH ARTICLE
Genetic characterization of cassava (Manihot esculenta Crantz)
genotypes using agro-morphological and single nucleotide
polymorphism markers
Kumba Y. Karim1,3 • Beatrice Ifie3 • Daniel Dzidzienyo3 • Eric Y. Danquah3 •
Essie T. Blay3 • Jim B. A. Whyte2 • Peter Kulakow2 • Ismail Rabbi2 •
Elizabeth Parkes2 • Lucky Omoigui2 • Prince E. Norman1,3 • Peter Iluebbey2
Received: 19 August 2019 / Revised: 15 November 2019 / Accepted: 24 November 2019 / Published online: 23 December 2019
 The Author(s) 2019
Abstract Dearth of information on extent of genetic (r = 0.99***), number of leaf lobe and root dry matter
variability in cassava limits the genetic improvement of (r = 0.30*), number of leaf lobe and starch content
cassava genotypes in Sierra Leone. The aim of this study (r = 0.28*), and height at first branching and plant height
was to assess the genetic diversity and relationships within (r = 0.45**). Findings are useful for conservation, man-
102 cassava genotypes using agro-morphological and sin- agement, short term recommendation for release and
gle nucleotide polymorphism markers. Morphological genetic improvement of the crop.
classification based on qualitative traits categorized the
germplasm into five different groups, whereas the quanti- Keywords Cassava  Genetic diversity  Morphological
tative trait set had four groups. The SNP markers classified traits  SNP markers
the germplasm into three main cluster groups. A total of
seven principal components (PCs) in the qualitative and
four PCs in the quantitative trait sets accounted for 79.03% Introduction
and 72.30% of the total genetic variation, respectively.
Significant and positive correlations were observed Cassava (Manihot esculenta Crantz) is a very important
between average yield per plant and harvest index root crop, containing high carbohydrate levels, used for
(r = 0.76***), number of storage roots per plant and har- human consumption, animal feed and industrial applica-
vest index (r = 0.33*), height at first branching and harvest tions (Sánchez et al. 2009). The starchy storage roots of
index (0.26*), number of storage roots per plant and cassava have become the most important source of dietary
average yield per plant (r = 0.58*), height at first branching energy in sub-Saharan Africa (SSA) as they provide more
and average yield per plant (r = 0.24*), length of leaf lobe returns per unit of input than any other staple crop (Fregene
and petiole length (r = 0.38*), number of leaf lobe and et al. 2000; Scott et al. 2000; Nassar 2005). Cassava is a
petiole length (r = 0.31*), width of leaf lobe and length of hardy plant that survives in poor soils with low fertility,
leaf lobe (r = 0.36*), number of leaf lobe and length of leaf relatively producing higher yields than other root and tuber
lobe (r = 0.43*), starch content and dry matter content crops (Temegne et al. 2015). However, cassava genotypes
respond differently to diverse environmental (soil, climate)
and biotic factors (Dixon et al. 2002).
& Prince E. Norman In Sierra Leone, dearth of information on the extent of
penorman2008@yahoo.com genetic variation within the breeding population of cassava
1 limits the development of superior cassava genotypes.Sierra Leone Agricultural Research Institute, Tower Hill,
Freetown PMB 1313, Sierra Leone Determination of the genetic variation in breeding popu-
2 lation facilitates identification of useful genetic divergenceInternational Institute of Tropical Agriculture,
Ibadan PMB 5320, Nigeria imperative for cassava population improvement. Genetic
3 divergence in breeding population is evaluated by geneticWest Africa Centre for Crop Improvement, College of Basic
and Applied Sciences, University of Ghana, markers (Andrade et al. 2017). Genetic markers such as
P.O. Box LG 30, Legon, Accra, Greater Accra, Ghana agro-morphological markers had been used frequently in
123
318 Physiol Mol Biol Plants (February 2020) 26(2):317–330
preliminary studies because they are fast and easy approach most abundant marker system in plant, animal, and
for assessing the extent of diversity among germplasm microorganism genomes and are considered as the new
(Asare et al. 2011). Some of these morphological traits generation molecular marker for various applications. The
revealed the true diversity as perceived by farmers (Mckey SNPs are useful in detecting and distinguishing specific
et al. 2001; Pinton and Emperaire 2001). Elias et al. (2001) genetic variations even in a low diversity species (Ferri
also reported that morphological traits have a heritable ge- et al. 2010). The use of SNPs has accelerated the pace of
netic variation. As knowledge in scientific research pro- genetic diversity research and gains in selection rather than
gressed, molecular markers were noted to unravel the using the conventional technique alone. Thus, the objective
genetic constitution and significance of traits through DNA of the present study was to assess the genetic diversity and
fingerprinting, gene link detection, identification of geno- relationships within cassava germplasm using agro-mor-
types, gene introgression, germplasm characterization, phological and single nucleotide polymorphism markers.
phylogenetic analysis, and indirect selection of agronomic
traits (Souza 2015; Andrade et al. 2017). Such knowledge
underpins the use of appropriate and reliable agro-mor- Materials and methods
phological descriptor and molecular markers for the eval-
uation of genetic diversity (Fukuda and Guevara 1998). Plant material, experimental design and plot layout
Quantitative and qualitative morphological traits have
been used for systematic identification of genotypes, spe- The trials were established in-field at the Njala Agricultural
cies and genera of some crops (Smykal et al. 2008). Research Centre (NARC) experimental site, southern
Qualitative traits are usually controlled by few genes with Sierra Leone in the 2015/2016 cropping season. Njala is
major effects. These traits are easily observable, thereby situated at an elevation of 50 m above sea level, 8 060 N
making differentiation and identification of genotype latitude and 12 060 W longitude. A total of 102 cassava
easier. Conversely, quantitative traits are controlled by genotypes comprising 82 white and 20 yellow accessions
many minor genes with complex inheritance. These traits were evaluated to determine the extent of genetic diversity
are more affected by environmental effects and develop- within the breeding population (Table 1). The experiment
mental stage of the crop. was laid out in a 6 9 17 alpha lattice design with three
Genetic diversity studies using morphological traits replications. Stem cuttings measuring 30 cm in length each
alone are sometimes limited by the environment and were planted on the crest of ridges at 1 9 1 m spatial
genotype by environment interaction effects (Collard et al. arrangement. No fertilizer or herbicide was applied. Hand
2005). These limitations may not permit the accurate weeding was done when necessary.
detection of duplicates by morphological classification
technique alone. Collard et al. (2005) reported that the use Agro-morphological data collection
of molecular markers may permit the detection of genetic
differences among closely related genotypes. Characteri- A total of 22 agro-morphological traits comprising 11
zation of accessions may, therefore, be more reliable if quantitative and 11 qualitative traits were evaluated
molecular markers are closely associated with morpho- (Table 2) based on the agro-morphological descriptor of
logical traits. Various DNA markers have been utilized to cassava described by Fukuda et al. (2010).
assess genetic diversity in cassava germplasm (Okogbenin Harvest index (HI) was calculated at harvest as the ratio
et al. 2012). These include restriction fragment length of fresh root yield to the total fresh biomass (weight of
polymorphism (RFLPs) (Beeching et al. 1993), random roots and weight of above ground biomass).
amplified polymorphic DNA (RAPDs) (da Silva et al. Starch extraction was done at harvest using a method
2015), amplified fragment length polymorphism (AFLP) described by Benesi (2005).
(Fregene et al. 2000), simple sequence repeats (SSRs) Starch content was calculated as:
(Asare et al. 2011) and single nucleotide polymorphism DSW
(SNP) markers (Kawuki et al. 2009). Of the above marker Starch contentð%Þ ¼  100FM
systems, SSR and SNP markers are among competitive
markers for diversity studies. However, microsatellites may where DSW is the weight of dried starch and FM is the
be limited by the presence of stutter bands that produce weight of fresh tuber.
quasi-scoring in ladders lacking prominent bands thereby Root dry matter content (RDMC) determination was
making scoring difficult (Park et al. 2009) and poor done at harvest by selecting three representative roots from
transferability across species (Grattapaglia and Kirst 2008). the bulk of roots harvested from 5 plants. Cassava roots
Single nucleotide polymorphisms are more easily assayed were washed and shredded into pieces. A standard measure
per locus compared to microsatellites. The SNPs are the of 100 g weight of the fresh samples was taken and oven
123
Physiol Mol Biol Plants (February 2020) 26(2):317–330 319
Table 1 Storage root flesh attributes of 102 cassava genotypes used for the genetic characterization study
Genotype Storage root flesh color Genotype Storage root flesh color Genotype Storage root flesh color
TR0991 White TR0188 White TR1523 White
TR1827 White TR1782 White TR0218 White
TR1036 White TR1069 White TR0043 White
TR1013 White TR0075 White SLICASS6 White
TR0038 White TR0575 White TR0142 White
TR0768 White TR1696 White TR0813 White
TR0519 White TR0127 White TR0971 White
TR0523 White TR0329 White TR0593 White
TR0092 White TR0864 White TR0589 White
TR1198 Yellow TR0302 White TR0345 White
TR1028 Yellow TR1005 White TR0135 White
TR0949 Yellow TR1162 White TR1436 White
TR0300 Yellow TR0159 White TR1326 Yellow
TR0356 White TR0006 White TR0454 White
TR1769 White TR0255 White TR1736 White
TR0024 White TR0435 White TR0724 White
TR0275 White TR0428 White TR0310 White
TR0694 Yellow TR1791 White TR0417 White
TR0912 White TR0288 White TR0120 White
TR0105 White TR1716 Yellow TR0164 White
TR0171 White TR0432 White TR0263 White
TR0488 White TR0297 Yellow TR0590 White
SLICASS4 White TR1167 White TR0382 Yellow
TR0746 White TR1175 White TR0453 White
TR0740 White TR0821 White TR1154 White
TR0204 White TR0657 White TR1097 White
TR1518 Yellow TR1035 Yellow TR0545 White
TR0064 White TR1618 Yellow TR1041 Yellow
TR0455 White TR0613 White TR0745 White
TR0779 White TR0358 White TR0017 White
TR0820 White TR1796 Yellow TR0591 White
TR0915 White TR1288 White TR1159 White
TR1824 White TR0667 White TR1776 White
TR0812 White TR0256 White COCO White
dried with forced drought oven. Samples were reweighed buffer and then placed in a 65 C water-bath for 25 min
again to obtain a constant weight after 72 h at 65–70 C with gentle shaking. The tube was removed from the water
(Fukuda et al. 2010). bath and allowed to cool for 5–10 min. Proteins and
polysaccharides were precipitated by adding 200 ll of ice-
Molecular characterization cold 5 M potassium acetate and then mixed by gentle
inversions (this was placed on ice for 20 min). About
DNA was extracted at the International Institute of Tropi- 350 ll chloroform:isomyl alcohol was added (24:1) to the
cal Agricultural Bioscience Laboratory IITA, Ibadan, content and mixed gently with continuous rocking and
Nigeria using the method proposed by Dellaporta et al. centrifuged at 4000 g for 10 min. This was followed by the
(1983) with a slight modification described by Rabbi et al. addition of RNase. The crude pellets were precipitated by
(2014). Freshly harvested apical leaves of about 200 mg of transferring the upper layer to a new tube. One volume
each accession were used. Grinding of the leaf samples was (400 ll) of ice-cold isopropanol was added and mixed
done in a 1.2 ml extraction tube using 400 ll extraction gently for about 2–3 min and then chilled in - 20 C
123
320 Physiol Mol Biol Plants (February 2020) 26(2):317–330
Table 2 Qualitative and quantitative traits used to characterize 102 cassava genotypes
SN Trait descriptor Trait Score code Sampling
acronym time
Qualitative traits
1 Color of leaf vein CLV 3 = green; 5 = reddish green in\ half of lobe; 7 = reddish green in[ half of lobe; 9 = all 6 MAP
red
2 Root taste RT 1 = sweet; 2 = intermediate; 3 = bitter 12 MAP
3 Cassava mosaic CMD 1 = no visible symptom of disease; 2 = mild; 3 = low; 4 = intermediate; 5 = high 6 MAP
disease
4 Color of root pulp CRP 1 = white; 2 = cream; 3 = yellow; 4 = orange; 5 = pink 12 MAP
5 Lobe margins LM 3 = smooth; 7 = winding 6 MAP
6 Ease of peeling EP 1 = easy; 2 = difficult 12 MAP
7 Leaf color LC 3 = light green; 5 = dark green; 7 = purple green; 9 = purple 6 MAP
8 Color of apical CAL 3 = light green; 5 = dark green; 7 = purplish green; 9 = purple 3 MAP
leaves
9 Root shape RS 1 = conical; 2 = conical-cylindrical; 3 = cylindrical; 4 = irregular 12 MAP
10 Shape of central SCL 1 = ovoid; 2 = elliptical-lanceolate; 3 = obovate-lanceolate; 4 = oblong-lanceolate; 6 MAP
leaflet 5 = lanceolate; 6 = straight; 7 = pandurate; linear-piramidal; linear-pandurate; linear-
hostatilobalate
11 External color of ECSR 1 = white or cream; 2 = yellow; 3 = light brown; 4 = dark brown 12 MAP
storage root
Quantitative traits
12 Number of storage NSR Count 12 MAP
roots/plant
13 Harvest index HI Derived estimate 12 MAP
14 Root yield per plant RYPP Derived estimate 12 MAP
15 Root dry matter RDMC Direct measurement 12 MAP
content (%)
16 Starch content (%) STC Direct measurement 12 MAP
17 Number of leaf lobe NLL Count 6 MAP
18 Petiole length PL Direct measurement using meter rule 6 MAP
19 Plant height PH Direct measurement using meter rule 12 MAP
20 Length of leaf lobe LLL Direct measurement using meter rule 6 MAP
21 Width of leaf lobe WLL Direct measurement using caliper 6 MAP
22 Height at first HFB Direct measurement using meter rule 12 MAP
branching
freezer for 10 min to enhance DNA precipitation. It was et al. 2011). The VCF file was filtered for missing value
then centrifuged at 4500 g for 20 min and the supernatant and polymorphic SNPs with quality parameter and a call
was carefully discarded. rate greater than 80%, depth[ 95%, and minor allele
frequency of 0.01. The SNPs with MAF values less than
SNP genotyping 0.01 and loci with more than 40% missing SNP marker
data were considered non-informative and were removed.
For SNP genotyping, about 50 ll concentrated DNA Of the 8600 SNPs subjected to filtering, 5600 informative
sample of each sample was sent to Cornell University for SNP markers were retained for genetic diversity study.
genotyping-by-sequencing analysis. The GBS was deter-
mined as described by Elshire et al. (2011) and sequenced Data analysis
at the Institute of Genomic Diversity at Cornell University
using the Illumina HiSeq 2500. The raw HapMap file from Qualitative and quantitative phenotypic data analysis
Cornell University was first converted to a Variant call
format (VCF) for the analysis using perl programming The genetic variation among the studied genotypes for
language and TASSEL 5.0 (Bradbury et al. 2007; Elshire agro-morphological traits was explored using multivariate
123
Physiol Mol Biol Plants (February 2020) 26(2):317–330 321
analysis technique. Multivariate analysis of the 102 9 11 Results
qualitative data matrix and 102 9 11 quantitative data
matrix comprising of principal component analysis (PCA) Frequency distribution of accessions according
were performed separately in SAS 9.4 software version. In to qualitative traits
the PCA, Eigen-values and load coefficient values were
generated from the data set. The relevance of trait contri- Frequency distributions of the qualitative traits are pre-
bution to the variation accounted by each principal com- sented in Figs. 1 and 2. Genetic variability was observed
ponent was based on the absulute eigenvector arbitrary among the 102 cassava accessions for all of the variables
cutoff value of 0.30 (Richman 1988). The PCA and cor- evaluated. The results showed that 53.9% of the accessions
relation matrices were used to determine the relationships exhibited light-green leaves, 42.2% had dark-green leaves
among the traits. The organization and structure of the and 3.9% had purple-green leaves (Fig. 1a). About 67% of
morphological variability were visualized using the the accessions had green leaf vein, 27% had reddish-green
Ascending Hierarchical Clustering (AHC) to plot a in more than half of lobe and 6% had reddish-green in less
dendrogram. than half of the lobe (Fig. 1b). Lobe margin of 52% of the
accessions was smooth, while 48% had winding lobe
Molecular data analysis margin (Fig. 1c). The shapes of the central leaflet of the
accessions were 2.0% linear pandurate, 27.5% linear
The genetic analysis package Power Marker version 3.0 pyramidal, 29.4% pandurate, 24.5% oblong-lanceolate,
(Liu and Muse 2005) was used to generate pairwise dis- 6.9% straight or linear, 2.9% lanceolate, 3.9% obovate-
tance-based hierarchical clustering. lanceolate and 2.9% ovoid (Fig. 1d). The accessions
exhibited 74.0% no symptom, 14.0% mild symptom,
10.0% moderate symptom, 2.0% severe symptom and 0%
Fig. 1 Percent distribution of a leaf color; b color of leaf vein; c lobe margins; d shape of central leaflet; and e cassava mosaic disease among
102 genotypes of cassava
123
322 Physiol Mol Biol Plants (February 2020) 26(2):317–330
Fig. 2 Percent distribution of a color of apical leaves; b external color of storage root; c ease of peeling; d color of root pulp; e root taste; and
f root shape among 102 genotypes of cassava
very severe symptom of cassava mosaic disease severity total variation where the second, third and the fourth PC
(Fig. 1e). axes with eigenvalues of 1.70, 1.35 and 1.14 accounted for
In terms of color of apical leaves, 84.3%, 12.7% and 15.43%, 12.24% and 10.38% of the total variation,
2.9% genotypes had light green, green purple and purple respectively. The fifth, sixth and seventh PC axes with
apical leaves, respectively (Fig. 2a). For external color of eigenvalues of 0.99, 0.97 and 0.83 accounted for 8.99%,
storage roots, 5.9% of the accessions had white or cream, 8.84% and 7.43% of the total variation, respectively.
17.6% light brown and 76.5% dark brown storage roots The first principal component with reference to its high
(Fig. 2b). Approximately 90.1% of accessions were easy to factor loadings was positively associated with traits such as
peel while 9.8% were difficult to peel (Fig. 2c). About root taste, color of root pulp, ease of peeling, and root
70.6% of the accessions had white root pulp, while 19.6% shape. The second PC was associated with leaf and storage
had cream root pulp and 9.8% had yellow root pulp root characteristics (root taste, leaf color, and color of
(Fig. 2d). About 79.4% of the accessions had sweet root apical leaves); the third PC was associated with external
taste, 14.7% were classified as intermediate and 5.9% had color of storage root, ease of peeling, color of leaf vein and
bitter root taste (Fig. 2e). The accessions comprised of shape of central leaf lobe, while the fourth PC was asso-
three root shapes including conical (3.9%), conical-cylin- ciated with traits related to storage root characteristics
drical (77.5%) and irregular (18.6%) (Fig. 2f). (color of root pulp, external color of storage root, and root
shape) color of leaf vein and cassava mosaic disease. The
Principal component analysis of qualitative fifth PC was associated with characteristics such as root
characters taste, cassava mosaic disease, root shape, lobe margin and
color of apical leaves, the sixth PC was also associated
The eigenvalues and percentage variations of the principal with external color of storage root, cassava mosaic disease
component analysis are presented in Table 3. Seven prin- and lobe margin and the seventh PC was also associated
cipal components that accounted for 79.03% of the total with storage (root color of root pulp and root shape) and
variation among the genotypes were identified. The first PC cassava mosaic disease.
axis with eigenvalue of 1.73 accounted for 15.76% of the
123
Physiol Mol Biol Plants (February 2020) 26(2):317–330 323
123
Table 3 Principal component analysis, eigenvalues and percentage variation of eleven qualitative traits of 102 cassava genotypes
Traits Prin1 Prin2 Prin3 Prin4 Prin5 Prin6 Prin7
RT 0.43 - 0.37 - 0.04 0.22 0.32 0.09 - 0.25
ECSR - 0.14 0.26 - 0.32 0.52 - 0.05 0.41 0.02
CRP 0.48 0.10 0.12 - 0.41 0.06 - 0.02 0.48
ES 0.47 - 0.22 0.33 0.29 0.12 0.15 - 0.18
CLV - 0.17 0.00 0.51 0.36 - 0.27 0.02 0.21
CMD - 0.18 0.13 0.19 0.33 0.67 - 0.31 0.44
RS - 0.41 - 0.12 0.21 - 0.30 0.36 - 0.14 - 0.46
LM - 0.21 - 0.18 0.01 - 0.28 0.30 0.78 0.27
LC 0.22 0.60 0.03 - 0.12 0.05 0.01 - 0.07
CAL 0.12 0.50 - 0.07 0.03 0.33 0.13 - 0.32
SCL 0.03 - 0.23 - 0.65 0.08 0.15 - 0.20 0.19
Eigenvalue 1.73 1.70 1.35 1.14 0.99 0.97 0.82
Proportion 15.76 15.43 12.24 10.38 8.99 8.84 7.43
of variance
(%)
Cumulative 15.76 31.19 43.43 53.81 62.80 71.64 79.07
variance
(%)
Values in bold represent significant traits in the various principal components, RT root taste, ECSR external color of storage root, CRP color of root pulp, ES ease of peeling, CLV color of leaf vein, CMD cassava mosaic disease,
RS root shape, LM lobe margins, LC leaf color, CAL color of apical leaves, SCL shape of central leaflet
324 Physiol Mol Biol Plants (February 2020) 26(2):317–330
Fig. 3 Dendrogram showing relationships among 102 genotypes of cassava classified by Ward method using eleven 11 qualitative agro-
morphological traits
Genetic relationship among 102 cassava genotypes similarity for the 11 qualitative traits ranged from zero to
using 11 qualitative traits one with a mean similarity of 0.10. The cassava genotypes
were grouped into five distinct clusters at 0.06 similarities.
The hierarchical classification of qualitative traits grouped Groups III, IV and V have a high number of genotypes with
genotypes into five classes almost with the same charac- 55, 22 and 12, respectively. Ten and 3 individuals were in
teristics as a function of the variables (Fig. 3). The genetic clusters II and I, respectively (Fig. 3).
123
Physiol Mol Biol Plants (February 2020) 26(2):317–330 325
Table 4 Probability values, means and coefficient of variation of quantitative traits of 102 cassava genotypes
Source HI AYPP PL LLL WLL RDMC NSR STC PH NLL HFB
Genotype \ .0001 \ .0001 \ .0001 \ .0001 0.9709 0.214 \ .0001 0.3192 0.7652 \ .0001 0.0098
Mean 0.45 1.56 13.5 10.6 4.4 30.9 7.5 23.9 149.1 5.8 58.9
CV 19.1 32.7 20.9 11.6 100.4 8.8 23.1 11.8 17.55 17.3 55.1
Significant at alpha = 0.05. HI harvest index, AYPP average yield per plant (kg), PL petiole length (cm), LLL length of leaf lobe (cm), WLL width of leaf lobe (cm), RDMC root dry matter
content (%), NSR number of storage roots (count), STC starch content (%), PH plant height (cm), HFB height at first branching (cm), NLL number of number of leaf lobe
Table 5 Correlation coefficients among 11 quantitative traits of 102 cassava genotypes
HI RYPP PL LLL WLL RDMC NSR STC PH HFB NLL
HI 1.00
RYPP 0.76*** 1.00
PL 0.01 0.11 1.00
LLL 0.14 0.23 0.38* 1.00
WLL - 0.19 - 0.01 0.23* 0.36* 1.00
RDMC 0.08 0.00 - 0.04 0.01 - 0.16 1.00
NSR 0.33* 0.58** 0.14 0.19 0.01 0.03 1.00
STC 0.08 0.00 - 0.05 0.00 - 0.16 0.99*** 0.05 1.00
PH 0.16 0.03 0.06 - 0.18 - 0.17 0.00 - 0.03 - 0.03 1.00
HFB 0.26* 0.24* - 0.04 - 0.27* - 0.21* 0.06 0.08 0.05 0.45** 1.00
NLL 0.05 - 0.03 0.31* 0.43** 0.19 0.30* 0.11 0.28* 0.07 - 0.03 1.00
Significant at alpha = 0.05. HI harvest index, RYPP root yield per plant (kg), PL petiole length (cm), LLL length of leaf lobe (cm), WLL width of leaf lobe (cm), RDMC root dry
matter content (%), NSR number of storage roots (count), STC starch content (%), PH plant height (cm), HFB height at first branching (cm), NLL number of leaf lobe
Mean values and correlation coefficients Representation of variables of quantitative traits
for the eleven quantitative traits
The result revealed that the four main components
The mean values for harvest index, root yield per plant, accounted for 72.30% of the total variation among the
root dry matter content, number of storage roots and starch genotypes. The first factorial plane contains 22.18% of the
content were 0.5, 1.6 kg, 30.9%, 7.5 and 23.9%, respec- variance. The variables that significantly correlated with
tively (Table 4). Significant and positive correlations were axis 1 are: harvest index (47%), root yield per plant (49%),
observed between root yield per plant and harvest index root dry matter content (30%), number of storage roots
(r = 0.76***), number of storage roots per plant and har- (40%), and starch percentage (30%). The variables that
vest index (r = 0.33*), height at first branching and harvest were significantly correlated with axis 2 are: petiole length
index (0.26*), number of storage roots per plant and root (37%), length of leaf lobe (51%), width of leaf lobe (47%)
yield per plant (r = 0.58*), height at first branching and and height at first branching (- 37%). The variables sig-
root yield per plant (r = 0.24*), length of leaf lobe and nificantly related to axis 3 are: root yield per plant
petiole length (r = 0.38*), width of leaf lobe and petiole (- 35%), root dry matter content (55%), starch content
length (r = 0.23*), number of leaf lobe and petiole length (55%) and number of leaf lobes (30%). The variables
(r = 0.31*), width of leaf lobe and length of leaf lobe significantly correlated with axis 4 are: petiole length
(r = 0.36*), number of leaf lobe and length of leaf lobe (40%), plant height (64%), height at first branching (40%)
(r = 0.43*), starch content and root dry matter content and number of leaf lobe (36%) (Table 6).
(r = 0.99***), number of leaf lobe and root dry matter
content (r = 0.30*), number of leaf lobe and starch content Genetic relationship among 102 cassava genotypes
(r = 0.28*), and height at first branching and plant height using 11 quantitative traits
(r = 0.45**) (Table 5). Conversely, significant and nega-
tive associations were noted between height at first Hierarchical classification of quantitative traits grouped
branching and length of leaf lobe (- 0.27*), and between genotypes into four classes almost with the same charac-
height at first branching and width of leaf lobe (- 0.21*). teristics as a function of the variables (Fig. 4). The genetic
123
326 Physiol Mol Biol Plants (February 2020) 26(2):317–330
Table 6 Principal component analysis, eigenvalues, percentage variation and cumulative variance of eleven quantitative traits of 102 cassava
genotypes
Traits Prin1 Prin2 Prin3 Prin4
HI 0.47 - 0.11 - 0.28 - 0.13
RYPP 0.49 0.02 - 0.35 - 0.23
PL 0.17 0.37 0.00 0.40
LLL 0.24 0.51 0.06 0.02
WLL - 0.05 0.47 0.03 0.10
RDMC 0.30 - 0.22 0.55 - 0.07
NSR 0.40 0.10 - 0.20 - 0.21
STC 0.30 - 0.21 0.55 - 0.10
PH 0.10 - 0.27 - 0.14 0.64
HFB 0.19 - 0.37 - 0.20 0.40
NLL 0.26 0.26 0.30 0.36
Eigenvalue 2.44 2.14 2.07 1.31
Proportion of variance (%) 22.18 19.42 18.83 11.87
Cumulative variance (%) 22.18 41.60 60.43 72.30
Figures in bold represent significant traits in the various principal components; HI harvest index, RYPP root yield per plant (kg), PL petiole length (cm), LLL length of leaf lobe (cm),
WLL width of leaf lobe (cm), RDMC root dry matter content (%), NSR number of storage roots (count), STC starch (%), PH plant height (cm), HFB height at first branch-
ing (cm), NLL number of leaf lobe (count)
similarity for the eleven quantitative traits ranged from fact that most of the genotypes exhibited white root pulp,
zero to one with a mean similarity of 0.10. Cluster I con- and dark brown external storage root. The above-ground
tains 19 genotypes, cluster II 12 genotypes, cluster III 46 leaf attributes of the studied genotypes were light green
genotypes and cluster IV 25 genotypes. apical leaves, light green leaf, green leaf vein, smooth lobe
margin and pendurate central leaflets. The leaf attributes
Clustering analysis using SNPs marker play key role in cultivar identification and are more
important for selection of cassava for the leafy veg-
The dendrogram showing clustering analysis of 96 cassava etable markets in Sierra Leone where cassava leaves are
genotypes based on 5600 SNP markers is presented in consumed. These findings concur with Agre et al. (2016)
Fig. 5. At similarity of 0.41, the result revealed three main who reported that farmers use the color of the leaves and
clusters. At similarity of 0.37, the accessions were further stems to identify their cassava cultivars.
divided into 7 sub-clusters. Cluster I consists of two sub- The principal component analysis is a powerful data
clusters: sub-clusters A and B. Sub-cluster A had two reduction technique utilized to reduce large number of
genotypes (TR0971 and TR0912) and sub cluster B con- correlated variables to a small number that is independent
tained 18 accessions. Cluster II consists of two sub-clus- and very useful. The PCA unraveled traits that contributed
ters: sub-clusters C and D. Sub-cluster C comprises of 6 most to the variation present in the cassava germplasm.
accessions; while sub-cluster D contains 22 accessions. The qualitative traits that contributed positively highest to
Cluster III consists of four sub-clusters: sub-clusters E, F, the first PCA include root taste, color of root pulp and ease
G and H comprising 5, 33, 3 and 6 accessions, respectively. of peeling. Findings of this study indicate the usefulness of
these traits for genotype identification and genetic diversity
studies in cassava. These are among key traits often con-
Discussion sidered relevant for selection of varieties for the genetic
improvement of the crop.
The analysis of qualitative morphological traits (root taste, The clustering based on similarity index of the qualita-
external color of storage root, color of root pulp, ease of tive traits in this study grouped the 102 cassava accessions
peeling, color of leaf vein, cassava mosaic disease, lobe into five clusters. Cluster I contained the accessions char-
margins, leaf color, color of apical leaves and shape of acterized by green apical leaves, cluster II grouped acces-
central leaflet) showed a significant variation among the sions having green apical leaves, smooth lobe of leaf
studied genotypes. Color was apparently the most repre- margin and resistance to cassava mosaic disease. Cluster III
sentative and the most distinctive trait possibly due to the was grouped based on ease of peeling, sweet root taste and
123
Physiol Mol Biol Plants (February 2020) 26(2):317–330 327
Fig. 4 Dendrogram showing relationships among 102 genotypes of cassava classified by Ward method using eleven 11 quantitative agro-
morphological traits
123
328 Physiol Mol Biol Plants (February 2020) 26(2):317–330
quantitative variables present yield and yield attribute traits
such as harvest index, average yield per plant, number of
storage root, root dry matter content and starch content that
may be integrated into a cassava breeding program. The
quantitative traits with highest positive contribution to
distinguishing genotypes in the first PCA included harvest
index, average yield per plant, number of storage roots,
root dry matter content and starch content. These are
among key traits often considered relevant for selection of
varieties and for the genetic improvement of the crop. The
cluster analysis of the 11 quantitative agro-morphological
traits also grouped the accessions into four clusters. Cluster
I accessions were characterized by high starch content, root
dry matter content and harvest index. Cluster II accessions
were characterized by high root dry matter content and
fresh storage root yield. Cluster III accessions exhibited
high starch content; whilst cluster IV accessions contained
high root dry matter content. The results generally indicate
the relevance of the above yield and yield attribute traits in
characterizing the genotypes. It also depicts the usefulness
Fig. 5 Dendrogram generated by UPGMA method of 96 cassava of the agro-morphological descriptor by Fukuda et al.
accessions based on SNP markers (2010) in identifying variability and reducing dimension-
ality in the traits set. In this study, the 11 qualitative and 11
conical cylindrical root shape. Cluster IV had light green quantitative trait sets sufficiently discriminated the 102
leaves and green apical leaves. Cluster V contained genotypes into distinct cluster groups. All accessions differ
accessions with dark brown external storage roots, light from each other in one or more traits with no detection of
green leaves and ease of peeling. In a similar study, Raghu duplicates, which suggest their usefulness in genotypic
et al. (2007) identified six distinct groups using 58 cassava differentiation and identification.
accessions. In this study, the first two principal components Findings of the molecular study revealed that 96% of the
explained 31.18% of the total cumulative variance for the 5600 SNP markers were polymorphic. The highest poly-
qualitative traits. This result is similar with those of Afonso morphic information content (PIC) value observed was
et al. (2014) who found 32.56% of the genetic variance in 0.17. The variation observed reflect the genetic constitution
the first factorial plane. It can also be explained by the fact of the accessions. In a previous study on cassava genetic
that the variance distribution is associated with the nature diversity using SNPs, Kawuki et al. (2009) reported PIC
and number of characters used in the analysis and focuses values of 0.228 in 74 cassava accessions. Moreover, in
on the first principal components. maize, Yang et al. (2011) reported a higher PIC value of
The analysis of the 11 quantitative traits revealed sig- 0.34 using 884 SNP markers. The variance in PIC values
nificant differences, seven of which had high coefficients of among these studies could be attributed to the number of
variation. The high coefficients of variation observed for genotypes and type of SNP markers used. Both the mor-
the examined characters indicated the presence of high phological and SNP markers established the uniqueness
heterogeneity within the population and therefore can be and variability within the cassava germplasm utilized in
exploited for breeding. Similar results for cassava were this study. The unique diversity in the cassava germplasm
obtained in Benin by Agre et al. (2015) where some suggests that the germplasm might possess genes, in high
averages were identical in cassava diversity study. In this frequencies, for adaptation in the studied area, whereas the
study, starch was positively and highly correlated with dry high genetic diversity is indicative of a high amount of
matter content indicating that starch content and dry matter additive genetic variance, needed for genetic progress in
content are closely related. Similar studies conducted at plant breeding. The high genetic variability also represents
CIAT and IITA have established that dry matter content a heterotic pool that provides an opportunity for the sys-
and starch content are closely correlated (r = 0.81) (IITA tematic exploitation of hybrid vigor in cassava. Although
1974; CIAT 1975). high diversity has been noted for African cassava germ-
The first four principal components analysis explained plasm (Lyimo et al. 2012), however, such diversity is still
72.30% of the overall variability in the quantitative anal- lower than those observed in Southern America cassavas
ysis. Principal components I, II and III obtained from (Hurtado et al. 2008; da Silva et al. 2015). Unlike the later
123
Physiol Mol Biol Plants (February 2020) 26(2):317–330 329
where farmers use seedling and vegetative propagation article are included in the article’s Creative Commons licence, unless
techniques (Siqueira et al. 2009; Mezette et al. 2013), indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons licence and your intended
farmers in Sierra Leone only propagate the crop using stem use is not permitted by statutory regulation or exceeds the permitted
cuttings. This study established the true-to-type genetic use, you will need to obtain permission directly from the copyright
identity and useful variability within cassava germplasm of holder.To view a copy of this licence, visit http://creative-
Sierra Leone needed for the genetic improvement and commons.org/licenses/by/4.0/.
conservation of the crop.
References
Conclusion Afonso SDJ, Ledo CAdaS, Moreira RFC, Silva SdeOe, Leal VDdeJ,
Conceição ALdaS (2014) Selection of descriptors in a morpho-
This study successfully determined the extent of genetic logical characteristic considered in cassava accessions by means
diversity within cassava breeding population of Sierra of multivariate techniques. IOSR-JAVS 7: 13–20
Agre AP, Dansi A, Rabbi IY, Battachargee R, Dansi M, Melaku G,
Leone using morphological and SNP markers. It also pro-
Augusto B, Sanni A, Akouegninou A, Akpagana K (2015) Agro
vides a data base for cassava breeders to make informed morphological characterization of elite cassava (Manihot escu-
decision for parental selection in a cassava breeding pro- lenta Crantz) cultivars collected in Benin. Int J Curr Res Biosci
gram based upon genetic diversity. The useful genetic Plant Biol 2:1–14
Agre AP, Badara G, Adjatin A, Dansi A, Bathacharjee R, Rabbi IY,
variability for storage root number, starch content, root dry
Dansi M, Gedil M (2016) Folk taxonomy and traditional
matter content and storage root yield that were identified management of Cassava (Manihot esculenta Crantz) Diversity
could be exploited for the genetic improvement of the crop in Southern and Central Benin. Int J Innov Sci Res 20:500–515
and its conservation. The color attribute of various quali- Andrade EKV, Andrade Júnior VC, Laia ML, Fernandes JSC,
Oliveira AJM, Azevedo AM (2017) Genetic dissimilarity among
tative traits studied contributed most to the differentiation sweet potato genotypes using morphological and molecular
of genotypes. The agro-morphological and SNP marker descriptors. Acta Sci Agron 39:447–455
techniques were complementary in distinguishing the cas- Asare PA, Galyuon IKA, Sarfo JK, Tetteh JP (2011) Morphological
sava genotypes. Both approaches should therefore be used and molecular based diversity studies of some Cassava (Manihot
esculenta Crantz) Germplasm in Ghana. Afr J Biotech
for genetic diversity studies of cassava.
10:13900–13908
Beeching JR, Marmey P, Gavalda MC, Noirot M, Haysom HR,
Acknowledgements The authors are grateful to the cassava breeding Hughes MA, Charrier A (1993) An assessment of genetic
team at the International Institute of Tropical Agricultural Nigeria diversity within a collection of cassava (Manihot esculenta
(IITA) for their technical support during the course of this study. Crantz) germplasm using molecular markers. Ann Bot
72:515–520
Author contributions All authors contributed to the study concep- Benesi M (2005) Characterization of Malawian Cassava Germplasm
tion and design. Material preparation, data collection and analysis for diversity, Starch extraction and its Native and modified
were performed by KYK, JBAW, PK, IR, EP, LO, PEN and PI. The properties. PhD thesis, Free State University, South Africa
supervision was done by BI, DD, EYD, ETB and JBAW. The first Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y,
draft of the manuscript was written by KYK and PEN, and all authors Buckler ES (2007) TASSEL: software for association mapping
commented on previous versions of the manuscript. All authors read of complex traits in diverse samples. Bioinformatics
and approved the final manuscript. 23:2633–2635
CIAT (1975) Centro International de Agricultural Tropical (CIAT)
Funding This research was funded by the International Development Cali, Colombia. Annual Reports, 1974
Research Centre (IDRC) under the IDRC/CORAF-WECARD/IITA Collard BCY, Jahufer MZZ, Brouwer JB, Pary ECK (2005) An
sub-grant agreement for developing capacity for Agricultural introduction to markers, quantitative trait loci (QTL) mapping
Research in Sub-Saharan Africa (PJ-2126) to pursue PhD in Plant and marker-assisted selection for crop improvement: the basic
Breeding at the West African Centre for Crop Improvement concepts. Euphytica 142:169–196
(WACCI), University of Ghana; and the West Africa Agricultural da Silva LI, Filho PSV, da Costa TR, Domingos ML, Goncalves-
Productivity Program Sierra Leone (WAAPP 1C SL) Grant Number: Vidigal MC (2015) Molecular characterization of traditional
IDA Grant H654-SL and Japan PHRD TF099510-SL. Assessions from the periurban region, Toledo, Western Parana,
Southern Brazil. J Glob Biosci 4:1268–1278
Compliance with ethical standards Dellaporta SL, Wood J, Hicks JB (1983) A plant DNA mini
preparation Version II. Plant Mol Biol Rep 1:19–21
Conflict of interest The authors declare that they have no conflict of Dixon AGO, Ngeve JM, Nukenine EN (2002) Genotype 9 environ-
interest. ment effects on severity of cassava bacterial blight disease
caused by Xanthomonas axonopolis pv. manihotis. Eur J Plant
Open Access This article is licensed under a Creative Commons Pathol 108(8):763–770
Attribution 4.0 International License, which permits use, sharing, Elias M, McKey D, Panaud O, Anstett MC, Robert T (2001)
adaptation, distribution and reproduction in any medium or format, as Traditional management of cassava morphological and genetic
long as you give appropriate credit to the original author(s) and the diversity by the Makushi Amerindians (Guyana, South Amer-
source, provide a link to the Creative Commons licence, and indicate ica): perspectives for on-farm conservation of crop. Euphytica
if changes were made.The images or other third party material in this 120:143–157
123
330 Physiol Mol Biol Plants (February 2020) 26(2):317–330
Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES, analysis and validation of resistance to cassava mosaic disease in
Mitchell ES (2011) A robust, simple genotyping-by-sequencing elite cassava genotypes in Nigeria. Crop Sci 52:2576–2586
(GBS) approach for high diversity species. PLoS ONE 6:e19379 Park Y-J, Lee JK, Kim NS (2009) Simple sequence repeat polymor-
Ferri L, Perrin E, Campana S, Tabacchioni S, Taccetti G, Cocchi P, phisms (SSRPs) for evaluation of molecular diversity and
Ravenni N, Dalmastri C, Chiarini L, Bevivino A, Manno G, germplasm classification of minor crops. Molecules
Mentasti M, Fani R (2010) Application of multiplex single 14:4546–4569
nucleotide primer extension (mSNuPE) to the identification of Pinton F, Emperaire L (2001) Cassava in the Brazilian Amazon:
bacteria: the Burkholderia cepacia complex case. J Microbiol varietal diversity and market. Genet Sel Evol 33:S491–S512
Methods 80:251–256 Rabbi I, Hamblin M, Gedil M, Kulakow P, Ferguson M, Ikpan AS, Ly
Fregene M, Bernal A, Duque M, Dixon AGO, Tohme J (2000) AFLP D, Jannink J-L (2014) Genetic mapping using genotyping-by-
analysis of African cassava (Manihot esculenta Crantz) germ- sequencing in the clonally propagated cassava. Crop Sci
plasm resistant to the cassava mosaic disease (CMD). Theor 54:1384–1396
Appl Genet 100:678–685 Raghu D, Senthil N, Saraswathi T, Raveendran M, Gnanam R,
Fukuda WMG, Guevara CL (1998) Morphological and agronomic Venkatachalam R, Shanmugasundaram P, Mohan C (2007)
descriptors for the characterization of manioc (Manihot escu- Morphological and simple sequence repeats (SSR) based finger
lenta Crantz). Cruz das Almas: Embrapa Cassava and Fruticul- printing of South Indian cassava germplasm. Int J Integrat Biol
ture, Documents 78 1:141–148
Fukuda WMG, Guevara CL, Kawuk R, Ferguson ME (2010) Selected Richman MB (1988) A cautionary note concerning a commonly
morphological and agronomic descriptors for the characteriza- applied eigenanalysis procedure. Tellus 40B:50–58
tion of cassava. International Institute of Tropical Agriculture Sánchez T, Salcedo E, Ceballos H, Dufour D, Mafla G, Morante N,
(IITA): Ibadan (Nigeria) Calle F, Perez JC, Debouck D, Jaramillo G, Moreno IX (2009)
Grattapaglia D, Kirst M (2008) Eucalyptus applied genomics: from Screening of starch quality traits in Cassava (Manihot esculenta
gene sequences to breeding tools. New Phytol 179:911–929 Crantz). Starch/Stärke 61:12–19
Hurtado P, Olsen KM, Buitrago C, Ospina C, Marin J, Durque M, de Scott GJ, Rosegrant MW, Ringler C (2000) Roots and tubers for the
Vincente C, Wongtiem P, Wenzel P, Killian A, Adeleke M, 21st century: trends, projections, and policy options. No 31, 2020
Fregene M (2008) Comparison of simple sequence repeat. Plant vision discussion papers, International Food Policy Research
Genet Resour 20:47–55 Institute (IFPRI). https://econpapers.repec.org/paper/fpr2020dp/
IITA (1974) Cassava annual report. IITA, Ibadan 31.htm
IITA (1990) Annual report. International Institute of Tropical Siqueira MVBM, Queiroz Silva JR, Brascan EA, Borges A, Pereira
Agriculture, Ibadan KJC, Pinto JG, Veasey EA (2009) Genetic characterization of
Kawuki R, Ferguson M, Labuschagne M, Herselman L, Kim DJ cassava (Manihot esculenta) landraces in Brazil assessed with
(2009) Identification, characterisation and application of single simple sequence repeats. Genet Mol Biol 32:104–110
nucleotide polymorphisms for diversity assessment in cassava Smykal P, Horacek J, Dostalova R, Hybl M (2008) Variety
(Manihot esculenta Crantz). Mol Breed 23:669–684 discrimination in pea (Pisum sativum L.) by molecular, bio-
Liu K, Muse SV (2005) Power marker: integrated analysis environ- chemical and morphological markers. J Appl Genet 49:155–166
ment for genetic marker data. Bioinformatics 21:2128–2129 Souza DCL (2015) Molecular techniques for characterization and
Mckey D, Emperaire L, Elias M, Pinton F, Robert T, Desmoulière S, conservation of medicinal and aromatic plants: a review. Rev
Rival L (2001) Local management and regional dynamics of Bras Plantas Med 17:495–503
varietal diversity of cassava in the Amazon. Genet Sel Evol Temegne NC, Ajebesone FN, Kuate AF (2015) Influence of soil
33:S465–S490 chemical composition on nutrient content and yield of cassava
Mezette TF, Blumer CG, Veasey EA (2013) Morphological and (Manihot esculenta Crantz, Euphorbiaceae) in two agro-ecolog-
molecular diversity among cassava genotypes. Pesq Agropec ical zones of Cameroon. Int J Biol Chem Sci 9:2776–2788
Bras 5:510–518 Yang J, Dong A, Peng Z (2011) Expression profiling of cassava
Nassar NMA (2005) Cassava: some ecological and physiological storage roots reveals an active process of glycolysis/glyconeo-
aspects related to plant breeding. An article published online genesis. J Integrat Plant Biol 53:193–211
with Gene Conserve
Okogbenin E, Egesi CN, Olasanmi B, Ogundapo O, Kahya S, Hurtado Publisher’s Note Springer Nature remains neutral with regard to
P, Marin J, Akinbo O, Mba C, Gomez H, de Vicente C, Baiyeri jurisdictional claims in published maps and institutional affiliations.
S, Uguru M, Ewa F, Fregene M (2012) Molecular marker
123