Food Bioscience 43 (2021) 101305

Contents lists available at ScienceDirect

Food Bioscience
journal homepage: www.elsevier.com/locate/fbio

Physico-chemical profiling of edible or sweet cassava (Manihot esculenta

Crantz) starches from Brazilian germplasm
Eduardo da Costa Nunes, PhD a, Virgilio Gavicho Uarrota, PhD b, *, Rodolfo Moresco, PhD c,
Marcelo Maraschin, PhD d
a
Santa Catarina Agricultural Research and Rural Extension Company (EPAGRI), Urussanga Experimental Station (EEUr), Rd. SC108-Km 353, 1563, SC, 88840-000,
Brazil
b
Pontificia Universidad Catolica de Valparaiso, Escuela de Agronomia, Calle San Francisco S/N, La Palma, Quillota, Chile
c
University of Itajaí Valley, School of the Sea, Science and Technology, Itajaí-SC, 88302-901, Rua Uruguai, 458, Brazil
d
Federal University of Santa Catarina, Morphogenesis and Plant Biochemistry Laboratory, Plant Science Center, Rod. Admar Gonzaga, 1346 - Itacorubi, Florianópolis -
SC, 88034-000, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: Edible or sweet cassava (i.e., with cyanide content ≤ 100 mg/kg, hereinafter referred to us as sweet cassava) is a
Cassava germoplasm staple food and source of energy in the feed industry and for the starch and ethanol industries. To expand the
Starch industrial uses of cassava, an increased emphasis in the search for value-added traits is urged and has not been
Amylose
reported. In this regard, the current study aimed to investigate the Physico-chemical and pasting properties of 40
Viscosity
Crystallinity
sweet cassava accessions from the Brazilian germplasm bank. The study prompts us to conclude that there is a
Chemometrics large genetic variability among the 40 genotypes. Genotype A0-109 presented higher starch content (61.74%)
and the AO-53 higher amylose content (21.65%). Genotypes 82 and A0-79 showed higher and lower swelling
power respectively. In general, the typical type “A” of starch crystallinity was observed and higher starch cris­
tallinity for genotype 54 and the lower for genotype 9. Starch from genotype AO-89 need higher pasting tem­
perature. A large proportion of the genotypic variance contained in the phenotypic variance was obtained for
pasting temperature (94%), peak viscosity (90%), breakdown, Final viscosity (88%) and amylose content (86%).
Genetic advance was higher for amylose content (25.61%) and swelling power (23.98%). Generally, genotypic
correlation coefficients were greater than phenotypic. This indicates that the genotypic component of variation
was the major contributor to total variation in the studied traits. Genotypes with higher swelling power are
suitable for use as thickeners and binding agents for food and non-food use. Those with lower setback are
desirable as gelling agents in refrigerated and frozen food products, for in natura consumption or semi-processed
products and can be prospected.

1. Introduction reversed to some degree when the International center of tropical

agriculture (CIAT), the International Institute of Tropical Agriculture
Cassava (Manihot esculeta Crantz) is an important energy source in (IITA) along with national programs were mandated to work on cassava
the diets of millions of people in the tropical and subtropical regions of on a global scale in 1970 (Ceballos et al., 2021; Hershey, 2017, p. 322).
the world. It is a key subsistence crop, and its industrial uses are recently Africa contributes 63.3% of the world production, followed by Asia
growing. According to the World Economic Forum (2021), cassava could (28%), Americas (8.6%) and Oceania (0.1%). The world top five pro­
potentially help alleviate the world’s hunger, provide economically ducers are Nigeria, the Democratic Republic of Congo, Thailand, Ghana
viable agriculture and even put an end to soil erosion and help farmers and Brazil (FAOSTAT, 2021). Although some increase in Africa pro­
bring once unproductive land back into use. Its minimal production in duction have been achieved recently, mainly due to expansion in crop
developed countries, have often meant that research investiment has area, cassava average productivity (9.1 t/ha) on the continent is well
long been far less than most important food crops. This situation was below the average cassava fresh root yield (21.5 t/ha) recorded in Asia

* Corresponding author.
E-mail addresses: eduardon@epagri.sc.gov.br (E. da Costa Nunes), uaceleste@yahoo.com.br (V.G. Uarrota), moresco@univali.br (R. Moresco), mtocsy@gmail.
com (M. Maraschin).

https://doi.org/10.1016/j.fbio.2021.101305
Received 26 April 2021; Received in revised form 2 July 2021; Accepted 6 August 2021
Available online 11 August 2021
2212-4292/© 2021 Elsevier Ltd. All rights reserved.
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

(Mbanjo et al., 2021). for 5 h for starch decanting. Subsequently, the supernatant was removed
Currently, it is widely believed that cassava’s greatest diversity is in and the deposited starch re-suspended in 2 L of distilled water and left in
Brazil. Cassava’s new variations are continually arising intentionally or this condition for 12 h (overnight), to obtain starch with higher purity.
not. It would be very difficult and expensive to collect and conserve a Then, the supernatant was completely drained and the starch mass
sample of every existing genetic variant of cassava or of the wild species. deposited at the bottom of the beaker collected, transferred to
What to conserve and how to identify and collect these materials are two aluminium trays and dried in an oven with air circulation at 45 ◦ C for 24
of the most fundamental bottlenecks in developing a conservation h. The dried samples were weighed and the extraction yield calculated
strategy as claimed by Hershey (2010, pp. pp1–99) and recently based on the dry weight. The starch samples of each access were then
reviewed by Ceballos et al., 2021. In order to establish a representative crushed using a coffee grinder, granulometry standardized to 60 mesh,
germplasm bank, it is necessary to collect cultivars and related wild stored in hermetic bottles and stored in a cool, dry place for further
species as means of their variability. Such variability goals must be analysis.
reached by an extensive profiling study of the genotypes present in the
germplasm banks. 2.3. Amylose content (%, w/w)
The starch fraction of cassava roots is basically formed by two
polysaccharides: amylose and amylopectin (Egharevba, 2019; Corne­ The amylose content was determined according to the protocol
jo-Ramírez et al., 2018; Kumar & Khatkar, 2017). Small contents of proposed by Megazyme® (Amylose/Amylopectin, 2006), using a com­
lipids and proteins are also found. The contents of these two macro­ mercial kit and based on the precipitation of amylopectin via conca­
molecules are variable according to the cassava botanical source and navalin (Con A). Concanavalin is a lectin (carbohydrate-binding
even between accesses of the same species, which will determine specific protein). It is a member of the legume lectin family. It binds specifically
characteristics of starch paste (Egharevba, 2019; Cornejo-Ramírez et al., to certain structures found in various sugars, glycoproteins, and glyco­
2018; Kumar & Khatkar, 2017). To meet the requirements of multiple lipids, mainly internal and non-reducing terminal α-D-mannosyl and
consumer markets it is necessary to characterize cassava starches by α-D-glucosyl groups. The concentration of amylose in the starch sample
means of their physicochemical properties. is estimated as the ratio of glucose oxidase plus peroxidase and 4-amino­
The Agricultural Research and Rural Extension Company of Santa antipyrine (GOPOD reagent) absorbance at 510 nm of the supernatant of
Catarina (EPAGRI), through the Urussanga Experimental Station, the Con A precipitated to that of the total starch sample.
located in Southern Brazil, maintains a cassava germplasm bank with
absorbance (Con A supernatant) 6.15
different genetic accesses (sweet cassava and industrial cassava species) Amylose (%) = * *100 (1)
Absorbance (Total starch aliquot) 9.2
that are conserved “on farm” by small farmers to support its cassava
breeding program. This germplasm has been evaluated based on
Where 6.15 and 9.2 are dilution factors for the Con A and Total Starch
eminently morpho-physiological and agronomic parameters. The
extracts respectively.
chemical composition and nutritional value of these accesses have not
been fully investigated and are of extreme importance for their potential
2.4. Swelling power
application. In this context, this study evaluated the physicochemical,
pasting and functional properties of the starch fraction of Forty (40)
Swelling power and solubility determinations were based on Leach
genetic accesses of sweet cassava deposited in that germplasm bank. It
et al.(1959), with modifications of the methods previously described by
was hypothesized that there is large genetic variability in cassava
Uarrota et al. (2013). Briefly, a starch sample (500 mg) was weighed into
starches that could be prospected.
a 50 ml graduated centrifuge tube. Distilled water was added to give a
total volume of 40 ml. The suspension was stirred sufficiently and heated
2. Materials and methods
at 95 ◦ C in a water bath for 30 min with constant stirring. The tubes were
cooled to room temperature and centrifuged (2500 rpm, 20 min). The
2.1. Selection of genetic accesses
supernatant was discarded and the residue mass (gel) was determined.
The swelling power of starches was estimated using equation (2).
Forty (40) genetic accesses of sweet cassava (Manihot esculenta
Crantz) were selected and harvested during two subsequent years at the SP(%) = Wg/Wm ∗ 100 (2)
Experimental Station of the Agricultural Research and Rural Extension
Company of Santa Catarina (Epagri) in Urussanga, Brazil. Genotypes Where: (SP-%) = Swelling power; (Wg) = weight of the gel; (Wm) =
were selected according to their economic, social and preservation weight of the sample with moisture corrected.
importance, genetic materials with pigmented pulp and some genetic
materials that had been kept on-farm by small family producers. The 2.5. Scanning electron microscopy (SEM)
genotypes were the 01; 03; 07; 09; 10; 11; 21; 23; 25; 31; 46; 54; 57; 59;
65; 74; 75; 79; 82; AO-13; A0-17; AO-40; AO-48; AO-49; AO-50; AO-53; 10 mg of the starch were mounted on double-adhesive carbon-coated
AO-54; AO-55; AO-72; AO-79; AO-80; AO-85; AO-89; AO-90; AO-94; tape on an aluminium stub. The sample was coated with gold using a
AO-97; AO-102; AO-109; AO-118 and AO-121. After harvest, the roots polaron E5001 SEM coating system. The coated sample was then viewed
(mixture of 3 Kg) were peeled, washed, packed in polyethylene pack­ under scanning electron microscopy (JEOL JSM-6390LV model, JEOL
ages, stored at − 20 ◦ C and transported to the laboratory where they Ltd., Tokyo, Japan) at 10 Kv and the micrographs were recorded with
were kept under the same conditions until the subsequent laboratory five replicates for each sample (Uarrota et al., 2013). The size of the
analyses. granules was measured directly from the micrographs with the magni­
fication of 2000x (5 measurements per image), with the aid of a ruler
2.2. Starch isolation and yield tool existing in the image capture software itself and/or with the aid of
the ImageJ® software. The forms were observed in micrographs with
Cassava root samples (~500 g) were ground with a manual grater magnifications of 500x and 2000x.
and then 2 L of distilled water were added and mixed by agitation. The
suspension was filtered using a system formed by two overlapping sieves 2.6. X-ray diffraction (XRD)
(60 mesh) and a filter element in a double layer. The retained biomass
was collected and washed with distilled water obtaining colourless The crystallinity of the starch granules was performed in an x-ray
filtrate. The filtrate was then transferred to a 5-L beaker and kept to rest diffractometer (X-ray Powder X’Pert PRO Multi-Purpose PanAnalytical,

2
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Netherlands). For this, starch samples were compacted in specific sup­ 3. Results and discussion
ports and scanted with Cu Kα radiation (λ = 1.5418Å), acting in the
angular range 2θ of 4-50◦ . Variation 2θ was used with a step of 0.3340 3.1. Starch yield and amylose contents
and a time interval of 10 s, operating in conditions of 45 kV e40 mA.
Data of starch yield and amylose contents followed the Gaussian
2.6.1. Relative crystalline index (CI) of starch fraction distribution (pvalue = 0.42 and 0.07 for starch and amylose respec­
The relative crystalline index (CI) was determined as proposed by tively) and is summarized in Fig. 1(a and b). Histograms with normal
Hayakawa (1997), with modifications, and the baseline of the dif­ distribution curves of both variables are presented in Fig. 1c-d respec­
fractogram peaks was defined from the profile obtained from a sample of tively. Starch ranged from 41.53% to 61.74%. One-way ANOVA and
cassava starch completely amorphized. The completely amorphized Tukey test of starch yield showed significant statistical differences be­
starch sample was obtained by mechanical crushing, using a high energy tween the genotype with higher starch yield, AO-109 (61.74 ± 2.84%),
ball mill (SPEX 8000) for 90 min. The CI was defined as the ratio be­ represented in red colour (Fig. 1a)compared with genotypes 7 and 74
tween the area of the crystalline region (CR) and the total area covered (41.55 ± 2.07 and 41.53 ± 1.39% respectively in blue colour) with
by the curve, composed of the area of CR and the area of the amorphic lower values. All other 37 genotypes were similar statistically.
region (Ar), and calculated according to equation Regarding the amylose contents (Fig. 1b), they ranged from 10.62 ±
1.53% (genotype AO-94) to 21.65 ± 0.30% (genotype AO-53). The
CI(%) = CR⁄ (CR + Ar) × 100 (3)
statistical analysis showed 03 significant different groups as represented
Origin software (version 7.0, Microcal, Inc., Northampton, MA, USA) by different colours in the graphic. Higher amylose content was
was used to calculate the integral of the amorphous (Ar) and crystalline observed for genotype AO-53, followed by AO-89, AO-72, AO-79 and
(CR) areas. AO-118 (red colour) and lower values for genotype 54, 23, 82, 7, AO-13,
AO-102 and AO-94 (blue colour).
Determination of the genetic variation in genetic accessions facili­
2.7. Pasting properties tates the identification of useful genetic divergence imperative for cas­
sava population improvement. Amylose content has a profound effect on
The rheological behaviour of cassava starches was studied using a starch functional properties (Karlstrom et al., 2016). Sanchez et al.,
Rapid ViscoAnalyser (RVA-4, Newport Scientific, Narrabeen, Australia) 2009 previously reported that the average starch content of the cassava
as previously reported by Uarrota et al. (2013). Briefly, starch samples accessions in the FAO germplasm collection is 84.5% (on a dry root
(28 g dry weight), suspended in deionized water (8.0% m/v), were basis) and the average amylose content is 20.7%, ranging between 15.2
subjected to different time-temperature profiles, e.g., heating from 48 ◦ C and 26.5%. Mtunguja et al. (2016) by assessing variation in physico­
to 95 ◦ C with a stirring speed of 160 rpm and incubation at 95 ◦ C/5 min. chemical, structural, and functional properties of root starches from
The viscosity was expressed in Pascal second (Pa.s). The breakdown is novel Tanzanian cassava landraces reported amylose contents from 11
the difference between the peak of viscosity and the minimum viscosity to 19% and starch from 74 to 80%. Previously, Mtunda (2009) have
during pasting, and setback is the difference between final and mini­ been reported starch contents ranging from 20.3 to 24.9% in 10 Tan­
mum viscosities during pasting. zanian genotypes. Contrarily, Nuwamanya et al., (2010) reported starch
values from 70.4 to 89.9% for Ugandan cassava cultivars. The average
starch content of our genotypes was 51.75% and amylose 16.08%. The
2.8. Fourier-transform infrared spectroscopy (ATR-FTIR) starch values were lower than those previously reported but differences
may be attributed to the differences in genetic sources, starch determi­
As previously reported by Uarrota et al. (2013), ATR-FTIR spectra of nation and environmental differences between the studied genotypes.
cassava starches were recorded in a Bruker IFS-55 (Model Opus v. 5.0, Recently, Karim et al., (2019) assessed the genetic diversity and re­
Bruker Biospin, Germany) spectrometer with a DTGS detector equipped lationships within 102 Sierra Leone cassava genotypes using
with a golden gate single reflection diamond attenuated total reflectance agro-morphological and single nucleotide polymorphism markers and
(ATR) accessory (45◦ incidence-angle). A background spectrum of the reported an average value of starch of 23.9 ± 11.8, lower values than
clean crystal was acquired and samples (100 mg) were spread and reported in our study. Njoku and Mbah (2020) also assessed starch
measured directly after pressing them on the crystal. The spectra were contents of 28 newly developed genotypes in Nigeria at the International
recorded at the absorbance mode from 4000 to 500 cm− 1 at the reso­ Institute of Tropical Agriculture (IITA) and reported an average value of
lution of 4 cm− 1. Five replicate spectra (128 co-added scans before 20.57 ± 0.79 through multivariate analysis such as path correlation
Fourier transform) were collected for each sample, in a total of 200 coefficients. When cassava is harvested too early, it often leads to a
spectra. For processing, the spectra were normalized, baseline-corrected reduction in yield, while delayed harvest leads to the development of
in the region of interest by drawing a straight line before resolution woody and fibrous tuberous roots and a reduction in starch content.
enhancement (k factor of 1.7) was applied using Fourier self deconvo­ Such observations were also reported by Benesi et al. (2007) who
lution (Vazhnova & Lukyanov, 2013)). The assumed line shape was studied the effect of genotype, season and location on starch content and
Lorentzian with a half-width of 19 cm− 1 (Čopíková et al., 2006). verified large genotypic variation on starch yield. The amylose values
reported in this study are similar to those reported by Mweta et al., 2008
(16.96–28.8%) for Malawian cassavas and those reported in the global
2.9. Data analysis and statistics analysis by Sanchez et al., 2009(15.2–26.5%).

The data were collected, summarized, submitted to normality test 3.2. Starch swelling power (SP), granule size, crystallinity and pasting
(Shapiro-Wilk test) and homogeneity of variances (Levene test) and then properties
to one-way analysis of variance (ANOVA). Where differences were
observed, means compared with the Scott Knott test (p < .05). All sta­ Swelling power represent evidence of interaction between the
tistics were performed in R software (R Core Team, 2021) by using amorphous and crystalline areas of starch. Furthermore, it is influenced
scripts developed by the research group. Data were also submitted to by amylose and amylopectin characteristics (Kusumayanti et al., 2015).
Pearson correlation, genetic variability analysis and Multivariate anal­ Fig. 2a shows the results of the swelling power of the present study.
ysis (Principal component analysis-PCA and hierarchical cluster One-way ANOVA showed significant genetic variability between the
dendrogram -HCA). cassava accessions evaluated that can be attributed to disparities in

3
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Fig. 1. (a) Starch yield of cassava genotypes and (b) Amylose content between cassava genotypes colored by groups. (c–d): Histogram with normal distribution curve
of starch yield of 40 genetic accessions evaluated.

bonding forces within the starch granule. Swelling power varied from =15.02, 16.99, 17.86 19.8, 22.9 which is typical of type A (Mtunguja
6.1 to 12.4%. Four different groups can be observed in Fig. 2a. Signifi­ et al., 2016). The relative crystallinity ranged from 32.8 to 44.6%.
cant Higher SP was observed for genotype 82 followed by 46, 79, 31 and Similar values were previously reported by Mtunguja et al., 2016
10. Lower SP was found for genotypes AO-79, AO-13, AO-50, AO-85, (36.1–41.4%). One-way ANOVA showed significant statistical differ­
AO-109, AO-80, AO-54 and AO-89 may be due to its higher amylose ences between the genotypes and 3 different groups, those with higher
content which impedes granule expansion. Sanchez et al., (2009) by crystallinity colored in red in Fig. 2b (i.e., genotypes 54, 57 and 82) and
evaluating 4050 cassava genotypes reported values ranging from 0.8 to lower crystallinity in blue (i.e., genotype 9, 46, 21 and 31). The crys­
15.5. Mtunguja et al. (2016) also reported values ranging from 8 to 15 talline type depends on the botanical source of the starch, i.e., A-type,
g/g in Tanzanian cassava genotypes. Starch swelling power depends on which is typically given by cereal starches; B-type by potato, amylo
the capacity of the starch molecule to hold water through hydrogen maize, canna, retrograded starch; and C-type by pulses, tapioca,
bonding during gelatinization. It measures the strength of interaction arrowroot and sago (Singh et al., 2006). A and B structures differ in the
between water molecules and glucan chains. As thermal energy in­ crystalline packing of the helices and the water content and C-structure
creases, the bonds between glucan chains relax and the granules absorb is a mixture of A and B unit cells and therefore intermediate between A
water and swell (Mtunguja et al., 2016). and B forms in packing density (Singh et al., 2006). Cassava starch may
Starch is a semi-crystalline material. The diffraction pattern of this be of the A-type crystallinity (Mtunguja et al., 2016).
study (Fig. 2c) showed prominent peaks with higher intensity at 2Ɵ Differences in granule size distribution may influence starch

4
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Fig. 2. (a) Swelling power (b) crystallinity index and (c) Wide-angle X-ray diffraction patterns of cassava starch.

functionality. Fig. 3a shows the results of the one-way ANOVA of the (Table 1). Significant higher values can be mentioned for genotype 9, 75,
granule size of this study. Significant 03 groups can be observed, those AO-121, 23, 79 and 21 while lower for AO-102, AO-118, AO-17 and
with large granule size (in red) and smaller granule size (in blue). Most AO-13. Peak viscosity ranged from 3.4 to 4.47 Pa s. Significant differ­
of the granules (Fig. 3b) were round and observed by scanning electron ences were also observed for breakdown and setback viscosities (Table 1
microscopy (SEM) and were similar to those previously seen in other and Supplementary Fig. 1). These parameters ranged from 1.98 to 2.98
studies by Mtunguja et al. (2016). The starch granule ranged from 6.6 to and 0.56 to 0.92 respectively. The observed values of peak viscosity,
17.2 μm. breakdown, setback are higher than those reported by Mtunguja et al.
Starch samples of cassava genotypes pasted from 62.6 to 70.1 0C (2016) in Tanzanian cassava genotypes.
(Table 1 and Supplementary Fig. 1). Significant statistical differences
were observed with means that there a large genetic variability between
the genotypes. 3 different groups were observed and genotype AO-89, 3.3. Genetic relationships among the 40 cassava genotypes via correlation
AO-102, AO-55 and AO-79 pasted with higher temperatures. and non-supervised multivariate analysis (PCA and phylogenetic trees)
Contrarily, the starch of genotypes 11, AO-54, 25, 23, AO-121, 59 and 9
pasted with lower temperature (See Supplementary Fig. 1). Sanchez The success of any breeding program greatly depends on the genetic
et al., 2009 while evaluating 4050 cassava genotypes reported values diversity available in the population. Understanding the magnitude of
ranging from 58 to 71 and Mtunguja et al. (2016) observed values from variability in crop species is pivotal since it provides the foundation for
66.4 to 69.6. Our results corroborate with those previously reported in selection. The genetic relationships among the 40 genotypes of cassava
the literature. Large variability was also observed for peak viscosity were firstly evaluated by Pearson correlation (p < .05). Table 2 sum­
marizes the results of such analysis. As it can be observed in the table,

5
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Fig. 3. Size of starch granule analytical starch granule morphology of the studied cassava genotypes. The image is representative of genotype 7. (A) 500x of
amplification and (B) 2000xshowing the starch granules.

positive relationships were found between breakdown with peak vis­ Heritability estimates give an insight into the extent of genetic con­
cosity, final viscosity with setback and peak viscosity, starch yield with trol to express a particular trait and phenotypic reliability in predicting
pasting temperature and swelling power with the final viscosity. its breeding value or also defined as a measure of precision of trials to
Contrarily, negative associations were observed between pasting tem­ compute the response to selection (Schimdt et al., 2019; Ene et al.,
perature with peak viscosity and breakdown, final viscosity with pasting 2016). High heritability indicates less environmental influence in the
temperature and the starch granule size with the final viscosity and observed variation (Schmidt et al., 2019; Ene et al., 2016). Since high
swelling power (Table 2). Results of the combined analysis of variance heritability does not always indicate a high genetic gain, heritability
presented in the Supplementary Table 1 showed that there are highly combined with genetic advance (Dyulgerova & Valcheva, 2014) should
significant differences among genotypes, indicating the presence of ge­ be used in predicting the ultimate effect of selecting superior varieties.
netic variability in the cassava germplasm studied. High ratios of the Genetic advance in this study (Table 3) was the higher for amylose
genotypic to phenotypic variance (Table 3) were observed for many content (25.61%) and swelling power (23.98%).
characteristics studied except for cristallinity index and granule size and Genotypic and phenotypic correlation coefficients for all pairs of
indicate that those characteristics were highly inherited and this appears then ten physic chemical characteristics are shown in the Supplementary
to show that further improvement through selection for these traits may Table 2. Generally, genotypic correlation coefficients were greater than
be effective. A large proportion of the genotypic variance contained in phenotypic correlation coefficients. This indicates that the genotypic
the phenotypic variance was obtained for pasting temperature (94%), component of variation was the major contributor to total variation in
peak viscosity (90%), breakdown and Final viscosity (88%) and amylose the studied traits. The highest negative genotypic correlation coefficient
content (86%). were observed between peak viscosity and pasting temperature (0.63),

6
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Table 1
Pasting characteristics (mean and standard deviation) of the 40 cassava genotypes evaluated. Viscosity is expressed in dynamic unit-Pascal second.
Peak Viscosity (Pa.s) Breakdown viscosity (Pa.s) Setback viscosity (Pa.s) Pasting Temperature (C)

Genotype Mean SD Mean SD Mean SD Mean SD

1 3.94 0.05 2.44 0.06 0.79 0.06 65.71 1,00

3 4.11 0.16 2.6 0.17 0.76 0.12 65.47 1.61
7 3.88 0.03 2.41 0.05 0.72 0.07 66.34 1.45
9 4.47 0.19 2.98 0.23 0.72 0.06 63.4 0.06
10 3.83 0.17 2.47 0.08 0.74 0.21 65.68 1.94
11 4.1 0.08 2.67 0.14 0.72 0.04 65.01 0.73
21 4.16 0.02 2.72 0.03 0.89 0.17 66.9 0.89
23 4.3 0.1 2.84 0.05 0.77 0.1 64.48 0.7
25 3.96 0.26 2.31 0.36 0.68 0.04 64.53 1.22
31 3.74 0.05 2.3 0.13 0.79 0.05 65.8 2.94
46 3.88 0.78 2.39 0.73 0.85 0.13 66.19 1.93
54 3.95 0.26 2.56 0.26 0.75 0.08 65.43 1.87
57 4.09 0.1 2.74 0.08 0.81 0.13 66.02 0.92
59 4.14 0.03 2.69 0.13 0.85 0.07 64.01 1.38
65 4,00 0.34 2.67 0.26 0.92 0.05 65.47 2.7
74 3.85 0.32 2.3 0.29 0.85 0.04 65.73 0.44
75 4.4 0.06 2.88 0.13 0.84 0.11 65.37 0.84
79 4.24 0.16 2.83 0.07 0.81 0.08 65.44 1.56
82 3.92 0.18 2.56 0.13 0.84 0.08 66.52 1.67
AO_102 3.68 0.14 2.38 0.09 0.64 0.03 69.22 0.54
AO_109 3.71 0.05 2.41 0.06 0.73 0.1 68.67 0.38
AO_118 3.6 0.06 2.42 0.02 0.78 0.08 67.97 0.74
AO_121 4.35 0.02 2.88 0.05 0.67 0.03 64.34 0.91
AO_13 3.4 0.67 1.99 0.75 0.74 0.06 65.76 1.33
AO_17 3.49 0.11 2.19 0.03 0.78 0.04 68.23 0.42
AO_40 3.88 0.01 2.49 0.02 0.68 0.09 68.31 1.12
AO_48 3.85 0.23 2.57 0.25 0.65 0.07 68.59 1.33
AO_49 3.88 0.09 2.47 0.21 0.84 0.08 66.64 0.62
AO_50 4.1 0.13 2.7 0.13 0.7 0.11 67.07 0.94
AO_53 3.77 0.21 2.32 0.32 0.56 0.04 68.27 0.51
AO_54 4.14 0.08 2.76 0.14 0.7 0.2 64.96 0.47
AO_55 3.86 0.18 2.42 0.1 0.68 0.02 69.06 0.46
AO_72 3.95 0.09 2.58 0.05 0.82 0.09 67.8 1.17
AO_79 3.75 0.12 2.46 0.03 0.71 0.01 68.83 0.57
AO_80 4.01 0.22 2.65 0.1 0.69 0.05 65.47 0.76
AO_85 3.72 0.05 2.49 0.01 0.8 0.18 66.78 0.74
AO_89 3.79 0.08 2.45 0.09 0.79 0.03 69.51 0.6
AO_90 3.78 0.1 2.49 0.13 0.73 0.08 66.57 0.53
AO_94 3.83 0.03 2.57 0.03 0.74 0.04 66.81 0.45
AO_97 4.02 0.15 2.71 0.11 0.76 0.06 65.38 0.95

Table 2
Pearson correlation coefficients for physico-chemical, pasting and functional properties of 40 cassava genotypes.
PV BR SB PT FV SY Amylose SP GS

Breakdown 0.91***
Setback 0.14 0.14
PastingTemp − 0.64*** − 0.48** − 0.25
FinalVisc 0.46** 0.15 0.64*** − 0.57***
Starch_yield − 0.07 0.13 0.09 0.33* − 0.31
Amylose − 0.02 − 0.00 − 0.15 0.27 − 0.13 0.18
SwellingPower 0.15 0.03 0.23 − 0.17 0.37* − 0.16 − 0.12
GranuleSize − 0.00 0.17 − 0.06 − 0.16 − 0.33* 0.02 − 0.30 − 0.39*
Crystallinity 0.06 0.11 − 0.07 0.03 − 0.11 0.08 − 0.24 − 0.03 0.11

Significance levels, p < .001, “***“, p < .01, “**“, p < .05, “*“. PV =peak viscosity; BR= Breakdown; SB= Setback; PT=Pasting temperature; FV= Final viscosity; SY =
Starch yield; SP= Swelling power; GS= Granule size.

final viscosity and pasting temperature (0.59) and swelling power and large genetic variability between the genotypes (Fig. 4A-B) and
granule size (0.58). Higher Positive genotypic correlations were also confirmed by the phylogenetic tree (Fig. 4C). The elbow cutoff suggested
found between peak viscosity and breakdown (0.92) and final viscosity 04 important components that in total contributed with 73.84% of data
with seatback (0.61). The magnitude of the genotypic and phenotypic variability. These four components contributed with 29.9, 18.37, 11.28
correlations and their utilization in the selection had been stated by a and 10.28% respectively. The two first components contributed 48.27%
number of researchers in the literature (Ene et al., 2016; Can & Yoshida, of data variability (Fig. 4A). Three different groups can be observed in
1999). Genotypic correlation coefficients offer a measure of the genetic the PCA. The contribution of the variables on each component can be
association between characteristics and may provide an important cri­ also found in Supplementary Fig. 2. Samples in the PC1 were mostly
terion of the selection procedures (Bernardo, 2020). influenced by peak viscosity, pasting temperature, final viscosity and
When data was submitted to non-supervised multivariate analysis, i. breakdown. PC2 was influenced by granule size, breakdown, swelling
e., principal component analysis (PCA) it can be observed that there is power and final viscosity. PC3 was influenced by amylose content,

7
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Table 3 Fig. 5a and Table 4). PCA on dimensions 2 and 3 separated the cultivars
Values of genotypic and phenotypic variances and ratios of genotypic variance 102 and 75 from the others due to the large presence of alkyl ethers and
to phenotypic variance (broad sense heritability) and genetic advance for the ten nitro groups (see Supplementary Fig. 4a). Finally, PCA on dimensions 2
cassava starch physico-chemical traits evaluated. and 3 showed differences of cultivars 102, 53, 57 and79 due to the
Trait δ2g δ2ph δ2g/δ2ph HE(bs) GA(%mean) presence of proteins, alkane and nitro groups (see Supplementary
SP 1.27 1.62 0.78 0.78 23.98 Fig. 4b). The squared loadings (i.e., the contribution of the variables) on
CI 1.78 3.7 0.48 0.48 4.79 the PCA performed showed a large contribution of peaks related to alkyl
SY 17.19 23.94 0.72 0.72 13.99 ether groups, amylose and amylopectin and proteins (Fig. 5b). Such
Amylose 4.68 5.47 0.86 0.86 25.61 differences and similarities were confirmed by hierarchical cluster
PV 0.055 0.061 0.9 0.9 11.7
BR 0.043 0.049 0.88 0.88 15.65
dendrogram where clearly can be observed that cultivars 75, 53, 102,
FV 0.014 0.016 0.88 0.88 10.85 57, 79 and 54 are different from other cultivars evaluated (Fig. 5c).
SB 0.005 0.007 0.71 0.71 17.41
PT 2.4 2.56 0.94 0.94 4.65 4. Conclusions
GS 1.31 5.26 0.25 0.25 9.26

PV =peak viscosity; BR= Breakdown; SB= Setback; PT=Pasting temperature; The genotypes had varied starch contents. Variability in amylose
FV= Final viscosity; SY = Starch yield; SP= Swelling power; GS= Granule size, content, swelling power, starch granule size, crystallinity and pasting
δ2g =genotypic variance, δ2ph =phenotypic variance, HE =heritability broad properties was also observed among the 40 cassava genotypes. Ac­
sense, GA= genetic advance as percentage of the mean. cording to the specific physicochemical and functional characteristics
they can be used in the food and non-food industry. Genotypes with
starch yield, breakdown and granule size of the starch and finally, PC4 higher swelling power are suitable for use as thickeners and binding
were highly contributed by starch yield, setback and relative crystal­ agents for food and non-food use. Those with lower setback are desirable
linity. Important variables can also be observed in Fig. 4A. The phylo­ as gelling agents in refrigerated and frozen food products. Genotypes
genetic tree (Fig. 4C) also confirms and highlights the genetic variability with higher peak viscosity may be difficult when mixing a starch paste.
between the cassava genotypes evaluated. Cassava genotypes with higher amylose content are potential (i.e. AO-
89, AO-118, AO-53, AO-72, AO-79) for use in the preparation of spe­
cific products that require greater crispness, resistance and low oil
3.4. ATR-FTIR spectra of cassava starches
adsorption (i.e. snacks). Products that require a more stable paste (i.e.
soups and sauces), needs starch with higher amylopectin content. A
ATR-FTIR spectra of cassava genotypes were summarized, the
large proportion of the genotypic variance contained in the phenotypic
spectra transformed from transmittance to absorbance, and then
variance was obtained for pasting temperature, peak viscosity, break­
normalized, baseline corrected and submitted to Savitzky–Golay first
down, Final viscosit and amylose content. Genetic advance was higher
derivative algorithm, before multivariate analysis in the ChemoSpec
for amylose content and swelling power.
package (Hanson, 2020). The typical chromatogram observed for all
Principal component analysis and phylogenetic tree also confirmed
cassava genotypes is presented in Supplementary Fig. 3 together with
such variability in the genotypes. As a whole, the results obtained can
the main peaks observed. Table 3 presents the main peaks observed in
also be used as support tools in the optimization of the selection of the
cassava starches and their corresponding assignment or chemical
groups. It can be observed the maximum peaks were observed for
Table 4
amylose, amylopectin (996/1018) and carboxylic groups - 2931
Typical chromatogram of ATR-FTIR spectra peak picking of cassava genotypes.
(Table 4). The found peaks were then submitted to multivariate analysis
(PCA and HCA) to see if there are differences or similarities between the Wavenumber (1/cm) Chemical group/Assignment

cassava genetic accessions evaluated. Fig. 5(a–c) and Suppl. Fig. 4 (a-f) 607, 671, 709, 765, Benzene, alkyl halide
summarizes the results. PCA analysis showed the importance of the first 928, 996, 1016, 1082, 1158, Amylose, Amylopectin, aryl alkyl ether, alkyl ether,
1242, 1372 alkyl amine, nitro group
3 components that contributed with 54.24% of the captured variance,
1418, 1424, 1456, 1562 Alkane
being 23.28, 16.14 and 15.49% for PC1, PC2 and respectively. PCA 1645, 1651 Proteins
performed on dimensions 1 and 2 showed clear separations of the cas­ 2931 Carboxylic acid
sava genotypes 53 and 75 from the other cultivars due to the large
Source: Andrade et al. (2020).
presence of alkane compounds and alkyl ether chemical groups (See

Fig. 4. (a) PCA biplot, (b) The squared loading of PCA and (c) Phylogenetic tree (“unrooted”) dendrogram of cassava genotypes.

8
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Fig. 5. (a) PCA biplot on the selected ATR-FTIR peaks (b) The squared loadings showing the contributions of the variables and (c) the cluster dendrogram of
cassava genotypes.

genetic improvement of sweet cassava. They demonstrate the impor­ Ceballos, H., Hershey, C., Iglesias, C., & Zhang, X. (2021). Theoretical and applied
genetics. In Fifty years of a public cassava breeding program: Evolution of breeding
tance of maintenance and conservation of these genetic materials in
objectives, methods, and decision-making processes. https://doi.org/10.1007/s00122-
germplasm banks due to the great genetic diversity presented, repre­ 021-03852-9
senting an important source of variability (genetic resources) with Copíková, J., Barros, A. S., Smidová, I., Cerná, M., Teixeira, D. H., Delgadillo, I., et al.
distinct characteristics to biomass, with enormous and diversified po­ (2006). Influence of hydration of food additive polysaccharides on FT-IR spectra
distinction. Carbohydrate Polymers, 63, 355e359.
tential for use and applicability, whether for natural use and/or in Cornejo-Ramírez, Y., Martínez-Cruz, O., Del Toro-Sánchez, C., Wong-Corral, F., Borboa-
specific industrial processes, due to the differential characteristics of Flores, J., & Cinco-Moroyoqui, F. (2018). The structural characteristics of starches
starch behaviour. and their functional properties. CyTA - Journal of Food, 16(1), 1003–1017.
Dyulgerova, B., & Valcheva, D. (2014). Heritability, variance components and genetic
advance of yield and some yield related traits in barley doubled haploid lines.
CRediT authorship contribution statement Turkish Journal Of Agricultural And Natural Sciences, Special Issue, 1(1), 614–617.
Retrieved 2 July 2021, from http://www.turkjans.com.
Egharevba, H. O. (2019). Chemical Properties of Starch and Its Application in the Food
Eduardo da Costa Nunes: Investigation, Methodology. Virgilio Industry, Chemical Properties of Starch, Martins Emeje, IntechOpen. https://doi.org/
Gavicho Uarrota: Conceptualization, Data curation, Formal analysis, 10.5772/intechopen.87777
Writing – original draft, Writing – review & editing. Rodolfo Moresco: Ene, C., Ogbonna, P., Agbo, C., & Chukwudi, U. (2016). Studies of phenotypic and
genotypic variation in sixteen cucumber genotypes. Chilean Journal of Agricultural
helped with the laboratory analysis. Marcelo Maraschin: Project Research, 76(3), 307–313.
administration, Resources, Funding acquisition, Supervision. Faostat. (2021). http://www.fao.org/faostat/en/#data/QC/visualize.
Hanson, B. A. (2020). ChemoSpec: Exploratory chemometricsfor spectroscopy. R package
version 5.3.2 https://CRAN.R-project.org/package=ChemoSpec.
Declaration of competing interest Hayakawa, K. (1997). Quality characteristics of hexaploid wheat (Triticumaestivum L.):
Properties of starch gelatinization and retrogradation. Cereal Chemistry, 74, 576–580.
Hershey, C. H. (2010). Background on “A global conservation strategy for cassava and wild
No potential conflict of interest was reported by the author(s). Manihot species”. pp1-.
Hershey, C. (2017). Achieving sustainable cultivation of cassava Volume 2 (1st Edition).
London: Burleigh Dodds Science Publishing, ISBN 9781351114318. https://doi.org/
Acknowledgements
10.4324/9781351114318
Karim, K., Ifie, B., Dzidzienyo, D., Danquah, E., Blay, E., Whyte, J., Kulakow, P., Rabbi, I.,
This research was fully supported by EPAGRI and Federal University Parkes, E., Omoigui, L., Norman, P., & Iluebbey, P. (2019). Genetic characterization
of Santa Catarina. Virgilio Uarrota thanks the Conicyt-Fondecyt of cassava (ManihotesculentaCrantz) genotypes using agro-morphological and single
nucleotide polymorphism markers. Physiology and Molecular Biology of Plants, 26(2),
3190055, Postdoctoral project and Pontificia Universidad Catolica de 317–330.
Valparaiso for research facilities. Karlström, A., Calle, F., Salazar, S., Morante, N., Dufour, D., & Ceballos, H. (2016).
Biological implications in cassava for the production of amylose-free starch: Impact
on root yield and related traits. Frontiers of Plant Science, 7.
Appendix A. Supplementary data Kumar, R., & Khatkar, B. (2017). Thermal, pasting and morphological properties of
starch granules of wheat (Triticum aestivum L.) varieties. Journal of Food Science &
Supplementary data to this article can be found online at https://doi. Technology, 54(8), 2403–2410.
Kusumayanti, H., Handayani, N., & Santosa, H. (2015). Swelling power and water
org/10.1016/j.fbio.2021.101305. solubility of cassava and sweet potatoes flour. Procedia Environmental Sciences, 23,
164–167.
Leach, H. W., McCowen, L. D., & Schoch, T. J. (1959). Structure of the starch granule I.
References
Swelling and solubility patterns of various starches. Cereal Chemistry, 36, 534e–544.
Mbanjo, E., Rabbi, I., Ferguson, M., Kayondo, S., Eng, N., Tripathi, L., et al. (2021).
Amylose/Amylopectin. (2006). Assay procedure for the measurement of the amylose and Technological innovations for improving cassava production in sub-saharan Africa.
amylopectin contents in starch. Wicklow: Ireland: Megazyme International Ireland Ltd. Frontiers in Genetics, 11. Omoregie Egharevba, H. (2020). Chemical properties of
Andrade, G., Medeiros Coelho, C., & Uarrota, V. (2020). Modelling the vigour of maize starch and its application in the food industry. Chemical properties of starch.
seeds submitted to artificial accelerated ageing based on ATR-FTIR data and Mtunda, K. J. (2009). Breeding, Evaluation, and selection of cassava for high starch content
chemometric tools (PCA, HCA and PLS-DA). Heliyon, 6(2), Article e03477. and yield in Tanzania. PhD thesis in plant breeding. South Africa: University of
Benesi, I., Labuschagne, M., Herselman, L., Mahungu, N., & Saka, J. (2007). The effect of Kwazulu-Natal.
genotype, location and season on cassava starch extraction. Euphytica, 160(1), Mtunguja, M., Thitisaksakul, M., Muzanila, Y., Wansuksri, R., Piyachomkwan, K.,
59–74. Laswai, H., Chen, G., Shoemaker, C., Sinha, N., & Beckles, D. (2016). Assessing
Bernardo, R. (2020). Reinventing quantitative genetics for plant breeding: Something variation in physicochemical, structural, and functional properties of root starches
old, something new, something borrowed, something BLUE. Heredity, 125(6), from novel Tanzanian cassava (ManihotesculentaCrantz.) landraces. Starch - Stärke,
375–385. 68(5–6), 514–527.
Can, N., & Yoshida, T. (1999). Genotypic and phenotypic variances and covariances in
early maturing grain sorghum in a double cropping. Plant Production Science, 2(1),
67–70.

9
E. da Costa Nunes et al. Food Bioscience 43 (2021) 101305

Mweta, D. E., Labuschagne, M. T., Koen, E., Benesi, I., & Saka, J. (2008). Some properties Schmidt, P., Hartung, J., Bennewitz, J., & Piepho, H. (2019). Heritability in plant
of starches from cocoyam(Colocasiaesculenta) and cassava (ManihotesculentaCrantz.) breeding on a genotype-difference basis. Genetics. 212(4), 991–1008. https://doi.
grown in Malawi. African Journal of Food Science, 2, 102–111. org/10.1534/genetics.119.302134
Njoku, D., & Mbah, E. (2020). Assessment of yield components of some cassava (Manihot Singh, V., Ali, S., Somashekar, R., & Mukherjee, P. (2006). Nature of crystallinity in
esculenta Crantz) genotypes using multivariate analysis such as path coefficients. native and acid modified starches. International Journal of Food Properties, 9(4),
Open Agriculture, 5(1), 516–528. 845–854.
Nuwamanya, E., Baguma, Y., Emmambux, M. N., & Rubaihayo, P. (2010). Crystalline and Uarrota, V. G., Amante, E. R., Demiate, I. V., Vieira, F., Delgadillo, I., & Maraschin, M.
pasting properties of cassava starchare influenced by its molecular properties. (2013). Physicochemical, thermal, and pasting properties of flours and starches of
African Journal of Food Science, 8–15. eight Brazilian maize landraces (Zea mays L.). Food Hydrocolloids, 30, 614–624.
R Core Team. (2021). R: A language and environment for statisticalcomputing. Vienna, Vazhnova, T., & Lukyanov, D. (2013). Fourier self-deconvolution of the IR spectra as a
Austria: R Foundation for Statistical Computing. URL https://www.R-project.org/. tool for investigation of distinct functional groups in porous materials: Brønsted acid
Sánchez, T., Mafla, G., Morante, N., Ceballos, H., Dufour, D., & Calle, F. (2009). sites in zeolites. Analytical Chemistry, 85(23), 11291–11296.
Screening of starch quality traits in cassava (ManihotesculentaCrantz). Starch/ World Economic Forum. (2021). https://www.weforum.org/agenda/2021/02/cassava
Stärke61, 12–19. -root-end-soil-erosion-deforestation/.

10