Received: 25 August 2021 Accepted: 5 January 2022 Published online: 8 April 2022

DOI: 10.1002/csc2.20715

Crop Science
ORIGINAL RESEARCH ARTICLE
Plant Genetic Resources

The World Vegetable Center Amaranthus germplasm collection:

Core collection development and evaluation of agronomic and
nutritional traits
Roland Schafleitner1 Ya-ping Lin1 Fekadu Fufa Dinssa2 Sognigbé N’Danikou2
Richard Finkers3 Ruth Minja4 Mary Abukutsa-Onyango5
Winnie Akinyi Nyonje5 Chen-yu Lin1 Tien-hor Wu1 Jeremiah Phanuel Sigalla2
Maarten van Zonneveld1 Yun-yin Hsiao1 Sanjeet Kumar1 Wan-jen Wu1
Hsin-I Wang1 Shou Lin1 Ray-yu Yang6
1 World Vegetable Center, P.O. Box 42, Shanhua, Taiwan 74199, Taiwan
2 World Vegetable Center, Eastern and Southern Africa, Duluti, PO Box 10, Arusha, Tanzania
3 Dep. of Plant Breeding, Wageningen Univ. and Research, P.O. Box 16, Wageningen 6700, The Netherlands
4 Mikocheni Agricultural Research Institute, P.O. Box 6226, Dar es Salaam, Tanzania
5 Jomo Kenyatta University of Agriculture and Technology, P.O. Box 62 000 – 00200, Nairobi, Kenya
6 Global Health Program, National Taiwan University, No. 17 Xuzhou Rd., Zhongzheng District, Taipei 100025, Taiwan

Correspondence
Roland Schafleitner, World Vegetable Abstract
Center, P.O. Box 42, Shanhua, Tainan, Amaranth (Amaranthus spp.) is an underutilized crop increasing in popularity as a
74199, Taiwan.
Email: roland.schafleitner@worldveg.org grain and as a leafy vegetable. It is rich in protein, minerals, and vitamins, and adapts
well to a range of production systems. Currently, the lack of improved cultivars
Assigned to Associate Editor Jason Gillman.
limits the use of the crop. Breeding-improved cultivars requires access to large
Funding information collections of amaranth biodiversity stored in genebanks. The task of searching such
German Federal Ministry for Economic vast collections for traits of interest can be eased by generating core collections,
Cooperation and Development (BMZ)
which display the diversity of large collections in a much smaller germplasm set.
commissioned by the Deutsche Gesellschaft
für Internationale Zusammenarbeit (GIZ) The World Vegetable Center amaranth collection contains around 1,000 accessions
through the Fund International Agricultural of 13 species; among them, there are 281 accessions of four species important for
Research (FIA), Grant/Award Number:
81219429
use as vegetable amaranth in Africa (A. cruentus, A. hypochondriacus, A. caudatus,
and A. dubius). Based on single nucleotide polymorphism (SNP) marker genotype
diversity, a core collection (CC) of 76 accessions, cultivars, and selections was
assembled. To a large extent, it represents the diversity of the whole collection. The
CC was evaluated for yield and nutritional parameters during the cool and warm
seasons in Tanzania and Taiwan and a pretest for variation of drought tolerance in

Abbreviations: CC, core collection; HQ, headquarters; SNP, single nucleotide polymorphism; WC, whole collection; WorldVeg, World Vegetable Center.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original
work is properly cited.
© 2022 The Authors. Crop Science published by Wiley Periodicals LLC on behalf of Crop Science Society of America.

Crop Science. 2022;62:1173–1187. wileyonlinelibrary.com/journal/csc2 1173

 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1174 Crop Science SCHAFLEITNER ET AL.

the CC has been performed. Cultivar Madiira 2, an improved cultivar developed

for vegetable production in Africa, outperformed all other tested cultivars in terms
of yield stability, but several CC accessions had higher yield, lower wilting score,
and higher nutrient content than Madiira 2. This indicates the core collection can be
used for further improvement of amaranth cultivars.

1 INTRODUCTION are preferred as leafy vegetables in Africa (Dinssa et al., 2020;

Shukla et al., 2004). Vegetable amaranth can be produced in
Several of the 40 to 50 species of the genus Amaranthus short rotation by thin broadcast sowing and uprooting of about
(Bonasora et al., 2013) are known as weeds, and Palmer ama- 3-wk-old plants, which avoids most pests and diseases. An
ranth is a particularly noxious one (Koo et al., 2018). But alternative production method is repeated harvest of larger
amaranth is also a food crop and is consumed in Africa and plants which increases yields per surface but makes the crop
much of the world. It is cultivated for its grains (or rather for more susceptible to pests and diseases that require specific
seeds, as amaranth is a pseudocereal), as well as for its leaves management methods (Nampeera et al., 2019) and generates
(vegetable use). Grain amaranth is an underutilized niche crop demand for improved cultivars. Small-scale farmers increas-
(Early, 1992; Sauer, 1967), whereas amaranth leaves are con- ingly seek dual use amaranth cultivars for multiple leaf col-
sumed in South and Southeast Asia, Brazil, Italy, Greece, lection during plant development and a final grain harvest
the Caribbean, and the Pacific Islands (Singh & Whitehead, (Hoidal et al., 2019). In collaboration with smallholder farm-
1996). Amaranth is particularly popular as a traditional veg- ers in East Africa, the World Vegetable Center (WorldVeg)
etable in sub-Saharan Africa (Ochieng et al., 2019; Oke, and partners have developed product profiles for improved
1980). The usual separation of Amaranthus species into grain cultivars. These cultivars were bred through selection from
and vegetable types (Joshi et al., 2018) does not hold in Africa, segregating accessions followed by cross breeding and were
where in addition to tetraploid A. dubius L., the leaves of released in several countries in sub-Saharan Africa (Dinssa
grain type amaranths such as A. hypochondriacus L., A. cau- et al., 2020). Current amaranth breeding at WorldVeg focuses
datus L., and A. cruentus L. are consumed as a vegetable. on rapid growth, taste, and nutritional quality of leaves of veg-
Due to its rapid growth and low production cost, amaranth etable and dual use types, broad spectrum resistance to leaf
is one of the cheapest leafy vegetables in tropical markets, yet diseases, and drought tolerance.
profitable for small-scale producers. As a C4 plant it grows Breeding success depends on access to crop biodiversity.
well during hot and dry summers, when other green vegeta- Amaranth biodiversity is maintained in several genebanks,
bles become rare on the market (Singh & Whitehead, 1996). and the largest amaranth ex situ germplasm collections
Amaranth leaves are rich in essential vitamins (pro-vitamin are held by the North Central Regional Plant Introduction
A, vitamin C) and minerals (calcium, Ca; iron, Fe; zinc, Zn; Station of USDA and by the National Bureau of Plant
Chakrabarty et al., 2018; Sarkar et al., 2020) that are often Genetic Resources in India (Genesys, 2021). The WorldVeg
low in local diets in Africa (Yang & Keding, 2009). The grain genebank currently holds a publicly accessible amaranth col-
contains about 15% of high-quality, gluten-free protein with a lection of more than 1,000 accessions of 13 species (https://
balanced content of essential amino acids for human nutrition. genebank.worldveg.org/#/?filter=v2z1YE963mb&p=0);
Leaves, like the grains, are also a good source of Ca, mag- however, this collection is too large for small breeding pro-
nesium (Mg), potassium (K), Fe, Zn (Bressani et al., 1993; grams to screen for traits of interest. Core collections (CCs)
Kachiguma et al., 2015; Mota et al., 2016), and antioxidants have been recognized as an efficient approach to promote the
(Sarkar et al., 2020). Amaranth is a neglected crop species use of new biodiversity in breeding and research (van Hintum
with high potential to contribute to the prevention of mal- et al., 2000). Core collections display a large fraction of the
nutrition, obesity, diet-related disorders, and hidden hunger diversity present in larger collections in a germplasm set small
(Pichop et al., 2014). enough to be efficiently screened for favorable agronomic
Amaranth breeding has mostly focused on grain types to and nutritional traits by smaller breeding programs with
improve yield, reduce plant height and seed shattering, and limited funding. Here we present the development of a CC
generate plants suitable for mechanical harvest (Das, 2016; for four amaranth species. The variation in yield and nutrient
Joshi et al., 2018). Vegetable amaranth breeding mainly tar- levels found in this germplasm is described in comparison to
gets cultivars of the species A. tricolor L. for the Asian current vegetable amaranth cultivars, with the aim to promote
market; much less effort has been made to improve A. cru- use of the CC to breed improved leafy and dual use amaranth
entus, A. dubius, A. caudatus, or A. hypochondriacus, which cultivars specifically for sub-Saharan Africa.
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHAFLEITNER ET AL. Crop Science 1175

2 MATERIALS AND METHODS

Core Ideas
2.1 Plant material and genotyping
∙ Amaranthus is a neglected crop with high nutri-
In 2018, out of 885 amaranth accessions available in the tional value but limited access to improved culti-
WorldVeg genebank, plants of 834 accessions were success- vars.
∙ Breeding-improved amaranth cultivars requires
fully grown in seedling trays. Leaf samples of two plants
per accession were lyophilized and DNA was extracted using access to trait variation.
∙ Core collections make it easier to search large
a column-based extraction kit (Favorgen). The DNA was
provided to Diversity Array Technology for genotyping by genebank collections for desirable traits.
∙ An amaranth core collection will support breeding
sequencing and bioinformatic analysis using the Diversity
Array Technology analysis pipeline, and around 76,000 sin- of vegetable and dual use (grain and vegetable) cul-
gle nucleotide markers (SNP) were obtained for 803 acces- tivars.
∙ The core collection contains accessions with supe-
sions and processed in Tassel 5 (Glaubitz et al., 2014). The
raw reads of this genotyping effort are available in NCBI Bio- rior yield, yield stability, and nutritional value.
Project PRJNA734966. The dataset was filtered for ≤50%
missing data and ≥0.25% minimum allele frequency. The
species annotations of the amaranth accessions of the World- WorldVeg breeding program (Supplemental Table S1). A set
Veg genebank were verified, and if necessary corrected (Lin of 1,860 markers with a minimum allele frequency of 1 and
et al., 2021). After taxonomic correction, in total 260 acces- 0% missing data in the whole collection (WC) and CC was
sions and 21 subaccessions (resulting from the separation selected and used to calculate the Shannon’s Diversity Index
of segregating accessions into subaccessions) belonged to A. and Nei’s expected heterozygosity in the WC and CC, as well
hypochondriacus, A. cruentus, A. caudatus, and A. dubius. as diversity in the various amaranth species of the WorldVeg
Characterization data for 11 quantitative and 22 qualitative genebank in PopGene (Yeh, 1999).
descriptors of the WorldVeg amaranth germplasm collected The representativeness of the CC on the phenotypic level
during seed regeneration over various years were obtained was assessed by comparing the trait values in the WC and the
from the WorldVeg genebank (Lin et al., 2021; Supplemental CC by T-test and F-test, respectively, and mean and variance
Table S1). Characterization data were missing for 39 (sub)- difference percentages, coincidence, and variable rate were
accessions; for 16 accessions data of more than one season calculated according to Hu et al. (2000). In addition, the per-
were available. centage of categories of qualitative data found in the WC and
maintained in the CC was calculated.

2.2 CC establishment, quality control, and

phylogenetic and population genetic analysis 2.3 Evaluation of agronomic and
nutritional traits
A CC representing the diversity of four amaranth species com-
monly used in Africa as vegetable and dual use types (A. The CC was evaluated over two seasons (warm and cool sea-
cruentus, A. hypochondriacus, A. dubius, and A. caudatus) sons) in the field at the WorldVeg Research Station in Arusha,
held by the WorldVeg genebank was established according Tanzania, and in the hot and rainy season at WorldVeg head-
to Odong et al., 2013. The genotypic data of the accessions quarters (HQ) in Taiwan (Shanhua, Tainan). In addition, a
were used to calculate a Euclidian distance matrix between drought tolerance prescreening was done in the summer sea-
accessions using the statistical analysis software R (https:// son in a screenhouse at WorldVeg HQ. Temperature data
www.r-project.org). An accession was randomly chosen, and were received from weather stations at WorldVeg Tanzania
all other accessions were grouped around it according to their and WorldVeg HQ, and during a gap in the records of the
genetic distance. This process was repeated for the remaining weather station at WorldVeg Tanzania, from worldweatheron-
accessions until 70 groups for the planned size of the CC of line.com (Supplemental Table S2). Improved amaranth cul-
about 70 accessions were established by iteratively adjusting tivars included in the core set were used as checks together
the chosen radius. From these groups, a set of representative with local cultivars as available (Supplemental Table S1). The
entries for the CC was created by minimizing the average dis- agronomic evaluation in Africa and Asia applied local man-
tance between each of the accessions and its nearest entry (A- agement conditions. In Africa, young leaves were harvested,
NE principle; Odong 2013). The core set was complemented whereas in Asia, after pinching the plants to stimulate side-
by adding improved cultivars such as Madiira 1 and Madi- shoot growth, young shoots were harvested. Under Asian con-
ira 2 (Dinssa et al., 2020) and breeding materials from the ditions, harvesting was started 16 d after transplanting vs. 30
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1176 Crop Science SCHAFLEITNER ET AL.

to 35 d after transplanting under African conditions. The har- For the dry down experiment, the CC set plus commercial
vesting intervals in Taiwan were shorter (about 7 vs. 14 d) and cultivars AM#53 and AM#74 were planted in plastic trays
a sixth harvest was performed during the cool season in Tai- with a soil depth of 23 cm in a randomized complete block
wan (Supplemental Table S2). Due to phytosanitary and logis- design with six plants per tray and four replicated plants per
tical reasons, not all accessions of the core set were available accession in a screenhouse on 22 July 2019 at WorldVeg HQ.
for the evaluations in Tanzania and Taiwan. The overlapping Irrigation was stopped in the drought treatment 21 d after after
set across all trials consisted of 64 entries of the 76-accession sowing (12 Aug. 2019). Wilting was scored on a scale from
CC (Supplemental Table S1). 0 (no wilting) to 6 (plant was dried out) 6, 9, and 12 d after
The seed for the warm season field trial in Arusha, Tanza- withholding irrigation.
nia (3.37˚ south, 36.8˚ east, elevation 1,235 m), was sown on Amaranth leaves sampled from the warm and cool sea-
4 Nov. 2019 and seedlings were transplanted on 27 Dec. The son trials at both locations were measured for mineral and
cool season trial was sown on 15 May 2020 and transplanted antinutrient content (oxalate). For measuring oxalate content,
on 15 June. Planting was done in a randomized complete block 100 mg of dried leaf powder was defatted three times in 1.8
design with three replications in plots of 24 plants arranged ml of hexane under shaking for 1 h. The suspensions were
in two rows with 60 cm between rows and 30 cm between centrifuged at 13,000 rpm for 5 min, the supernatant was dis-
plants. The area planted and harvested per plot comprised carded, the pellets were rinsed and then suspended in 10 ml of
6.48 m2 . Five harvests at 11- to 16-d intervals were performed distilled water, and 1.5 ml of the suspension was centrifuged
(Supplemental Table S2). The data collected during harvest at 13,000 rpm for 5 min and filtered through a membrane with
comprised number of leaves harvested per plot at each har- 0.45-μm pore size. Twenty μl of the filtered extract was sep-
vest, leaf yield per plot (g) at each harvest, leaf length and arated on a high performance liquid chromatography (HPLC;
width (cm) from 10 randomly picked leaves from 10 random Waters Alliance 2695) using an ICSep ICE-ORH-801 organic
plants at first and last harvest, and number of branches per 10 acid column (0.65 by 30 cm, Transgenomic) and oxalate was
random plants per plot at maturity stage. For nutrient analysis, quantified in comparison to a commercial standard using a
at least 300 g of leaves were randomly sampled from the mate- photo array detector (Waters). The running conditions were
rial harvested from multiple plants, weighed, and oven-dried. set at 35 ˚C using isocratic HPLC with 0.01 N H2 SO4 at 0.8 ml
Weight was taken after harvest and after drying. The dried min−1 . Calcium, Fe, and Zn analysis was performed by ICP-
samples were shipped to WorldVeg HQ for nutrient analysis. OES. For this purpose, 0.2 g of dried sample powder was
The yield trial during the warm season at WorldVeg HQ loaded into digestion tubes, 5 ml of H2 SO4 was added and
in Taiwan (23.12˚ north, 120.3˚ east, elevation 12 m) was on the following day the samples were heated to 300 ˚C for 2
performed under field conditions in a randomized complete h, removed from the digestion tubes, and cooled down to 150
block design with two replications. The commercial cultivars ˚C. Then 2 ml of 30% (v/v) H2 O2 was added, and the samples
AM#53 and AM#74 and some WorldVeg genebank acces- were incubated in the digester at 300 ˚C for 1 h, cooled to 40
sions of interest for breeding were used as additional checks ˚C, filled up to 50 ml with distilled water, and the elements
(Supplemental Table S1). The seed was sown on 21 May 2019 were determined on an 8000 ICP-OES (PerkinElmer).
in seedling trays in the greenhouse and seedlings were trans-
planted to the field on 15 July into double rows in 1.5- by
3.5-m plots, with 14 plants per plot. The plants were pinched 3 RESULTS
on 29 July. Five harvests at intervals of six to nine days
(Supplemental Table S2) were performed with a sickle. For 3.1 Core collection
comparisons of the yield with those of the other trials, the
lower number of plants per plot (10 vs. 20) was taken into Genotyping by sequencing resulted in 76,420 SNP markers
consideration. for 803 accessions of 13 species. Filtering the SNP sites for
The CC set tested during the warm season was also evalu- maximum missing data of ≤50% and a minimum allele fre-
ated under spring (cool season) conditions at WorldVeg HQ, quency of ≥0.25% resulted in 27,540 polymorphic markers
using the same field layout as the warm season trial. Instead of for the total genebank collection of 13 species. These markers
14 plants per accession, 24 plants per plot were grown, using were used to perform a diversity analysis on 281 accessions
the same spacing as the warm season trial. Commercial cul- of the four Amaranthus species mainly used for breeding-
tivars AM#53 and AM#74 and WorldVeg breeding materials improved vegetable and dual use cultivars in Africa: A. cru-
were used as checks (Supplemental Table S1). The seed was entus, A. hypochondriacus, A. caudatus, and A. dubius, to
sown on 11 Feb. 2020, seedlings were transplanted on 5 Mar., generate a CC of 70 accessions representing a maximum of
the plants were pinched on 17 Mar., and six harvests were per- the genetic diversity for this group. The CC consisted of a
formed (Supplemental Table S2). Material from the second similar proportion of A. hypochondriacus and A. cruentus
harvest was used for mineral and oxalate analysis. than the whole collection but contained a larger proportion of
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHAFLEITNER ET AL. Crop Science 1177

TA B L E 1 Quality assessment of the core collection (CC) based on quantitative phenotypic characteristics

Descriptor Am110 Am220 Am240 Am250 Am290 Am300 Am400 Am410 Am420 Am460 Am540
t-test .00** .91 .02* .11 .23 .60 .74 .54 .43 .92 .85
Mean WC 2.80 127.89 42.38 22.84 16.77 9.37 46.59 27.20 16.41 14.56 0.42
CC 2.52 128.82 33.07 18.95 17.66 9.19 47.43 26.18 15.22 14.73 0.43
Range WC 12 280.8 137.5 112 33.8 26.55 117 68.61 42.3 38.8 1.36
CC 5 264.6 107.4 73.3 20.7 12.8 85 48 39.4 38.3 0.99
F-test .00** .82 .13 .07 .37 .02* .81 .01** .92 .68 .33
CV WC .51 .48 .78 .83 .36 .34 .41 .56 .71 .61 .74
CC .37 .46 .86 .83 .31 .28 .39 .44 .75 .63 .65

Note. CV, coefficient of variation; WC, whole collection.

*Significant at the .05 probability level.
**Significant at the .01 probability level.

A. caudatus (16 vs. 8%) and a smaller proportion of A. were noted in the WC of the four amaranth species, and 67
dubius (18 vs. 29%), reflecting the intraspecific diversity of (80%) of these categories were maintained in the CC. Cate-
the germplasm. The CC was completed with important culti- gories absent in the CC concerned very slow germination rate
vars and selections, resulting in a CC of 76 accessions. The and the presence of spines in the leaf axils. There was less
representativeness of this CC was verified by analyzing pop- diversity of terminal inflorescence shapes and leaf, petiole,
ulation genetic parameters, and on the phenotypic level, by stem, and seed pigmentation in the CC than in the WC.
comparing the trait diversity of 11 quantitative and 22 quali- Compared with the WC and CC, the seven cultivars and
tative descriptors in the WC and the CC. selections included in the CC, on average, had a shorter ger-
Conservation of the allelic diversity in the core collection mination period (Am110), shorter mean length of basal lat-
was expected, as the method to generate the CC was based on eral branches (Am240), and shorter axillary inflorescences
selecting accessions representing the genetic diversity in the (Am460) than the mean of the CC, but were on average higher
WC. Population genetic analysis confirmed that the reduction (Am220), had larger leaves (Am290), were flowering later
of the collection size from 281 accessions to a CC comprising (Am400), had longer terminal inflorescences (Am420), and a
about 30% of the WC resulted in similar Shannon’s Diversity larger 1,000 seed weight (Supplemental Table S1). The aver-
Index (WC: .22, CC: .25) and Nei’s expected heterozygosity ages of the top lateral branch length (Am250), leaf width
(WC: .15, CC: .14), suggesting the genetic diversity of the WC (Am300), and terminal inflorescence stalk length were in the
was well maintained in the CC. range of the averages of the CC. For traits with higher means
A major indicator for the quality of a CC is the repre- in cultivars, CC accessions with even larger trait values than
sentation of the phenotypic diversity of the WC. The dif- the greatest value in the cultivars were available, except for
ference of the phenotypic means between the WC and CC larger leaves (Am290). However, there were accessions with
was insignificant (p = .93). Values were significantly dif- still larger leaves available in the WC (Supplemental Table
ferent between WC and CC for only two (Am110 - days S1).
to germination and Am240 - mean length of basal lateral
branches) out of the 11 quantitative traits (Table 1). The
coincidence and variable rates amounted to 73 and 93% and 3.2 Agronomic evaluation of the CC
were in the acceptable range for core collections (Hu et al.,
2000). According to F-test, there was a significant decrease The CC was evaluated for yield in the warm and cool sea-
of variance for three quantitative traits in the CC: number of sons in Arusha, Tanzania, and in Shanhua, Taiwan. The tri-
days to germination (Am110), leaf width (Am300), and ter- als were done under local cultivation conditions and local
minal inflorescence stalk length (Am410; Table 1). Acces- harvesting methods were applied. The warm season trial in
sions with a longer germination period (>6 d), narrower and Africa was not affected by pests and diseases, whereas in the
wider leaves (<3.9 cm and >16.7 cm), as well as longer ter- cool season, about one third of the plots were very mildly
minal inflorescence stalk (>49 cm) were not represented in affected, 4% of the plots were mildly affected by the insect
the CC. Variation in plant height (Am220), terminal lateral pest Hymenia recurvalis, and one fifth of the plots were very
inflorescence length (Am420), and length of lateral inflores- mildly affected by an unidentified fungal disease. No diseases
cence (Am460) were particularly well maintained in the CC. or pests were recorded in the trials in Taiwan. The differ-
From the 22 qualitative descriptors in a total of 93 categories ences of the minimum, maximum, and average temperatures
describing the form and color of plant organs, 84 categories between the warm and cool season trials in Tanzania were
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1178 Crop Science SCHAFLEITNER ET AL.

F I G U R E 1 Leaf yield per plot (20 plants) in Arusha, Tanzania (a) and World Vegetable Center headquarters in Taiwan (b). The highest yields
were obtained in Tanzania during the warm season and in Taiwan during the cool season

3, 5.6, and 2.9 ˚C, respectively. In Taiwan, the differences T A B L E 2 Correlations between yields at Harvests 1–5 (H1–H5)
between the minimum, maximum, and average temperatures with leaf length and width
in the cool and warm season trials were 7.8, 4.8, and 6.3 ˚C Season H1 H2 H3 H4 H5
(Supplemental Table S2). Among all trials in this study, the Arusha warm season
cool season trial in Taiwan and the warm season trial in Tan-
Leaf length .13* .1 .16* .35** .58**
zania showed the most similar temperature profiles. The min-
Leaf width .19* .12 .16* .39** .52**
imum temperatures in the cool season trial in Taiwan were on
Arusha cool season
average 0.9 ˚C warmer than in the warm season trial in Tanza-
Leaf length −.01 .41** .19** .23** .12
nia, whereas the maximum and average temperatures were on
average 2.2 and 0.7 ˚C higher in the warm season in Tanzania Leaf width .01 .32** .08 .047 .09
than in the cool season in Taiwan, respectively. *Significant at the .05 probability level.
Leaf yield varied among entries, seasons, and locations, **Significant at the .01 probability level.

but was significantly positively correlated between all trials

except for the warm season trial in Taiwan (Figure 1a and 1b; all trials. The best performers in Tanzania in terms of leaf
Table 2). Largest yields were achieved in the cool season in yield over the warm and cool seasons were cultivar Madi-
Taiwan and in the warm season in Tanzania, whereas during ira 2 (VI060470), accession VI061497, and selection IP-2
the warm season in Taiwan the yields were lowest. Average AVAM1602 (VI060472). Also in Taiwan, when consider-
yield reduction of the germplasm set in the Taiwanese warm ing both seasons, accession VI061497 was among the best
season was 73% compared with the cool season. In Tanza- performers, together with VI050458 and VI044475, whereas
nia, the yield reduction in the cool season compared with Madiira 2 ranked eighth for combined yield of both seasons.
the warm season was 48%. The largest yields across environ- Under the hot conditions of the warm season trial in Taiwan,
ments and seasons were obtained with accession VI044475 accessions VI044475 and VI050128 yielded best, with about
(A. dubius), selection IP-2 AVAM1602 (VI060472. A. cru- 50% of the yield of the cool season trial. Under cold conditions
entus), VI061497 (A. dubius), and VI050128 (A. hypochon- of the cool season trial in Tanzania, selection IP2 AVAM1602
driacus). Cultivar Madiira 2 (VI060470, A. cruentus) was (VI060472) yielded even more than in the warm season,
under the best performers and ranked sixth for yield across and cultivar Madiira 2 (VI060470) had only 22% yield loss
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHAFLEITNER ET AL. Crop Science 1179

F I G U R E 2 Average yields per harvest in the warm and the cool seasons in Tanzania (Arusha) and Taiwan in g plant−1 . Harvesting time refers
to the first to the sixth harvest (Supplemental Table S2)

compared with the warm season trial. Local Taiwan cultivars Taiwan. Overall, average Fe contents were higher in the trials
KY-white and KY-red had low yields in both the warm and in Tanzania than in Taiwan. Average Zn contents were similar
cool season trials in Taiwan (Figure 1b). across regions and seasons, but the variation in the germplasm
Yields of subsequent harvests in Tanzania increased over was the highest in the cool season in Tanzania (Table 3).
time during the warm season and only dropped at the fifth The highest Zn concentrations were found in cultivar Simon’s
harvest (Figure 2). During the cool season in Tanzania, the Farm (VI062433) across all environments. Ca contents were
yields per harvest were lower than in the warm season, and on average higher in the cool than in the warm season at
for many accessions no harvest could be obtained at the fifth both locations. The average concentration of oxalate was the
harvest. In Taiwan, all harvests during the warm season were lowest during the warm season in Taiwan, but significantly
much smaller than in the cool season. In the cool season of higher during the cool season (p < .01), whereas in Arusha
Taiwan, a sixth harvest was possible increasing the yields per the oxalate concentration difference across seasons remained
plant per season (Figure 2). insignificant. Correlations between yield and oxalate accumu-
In Tanzania, during the warm season, correlation between lation were significant in the cool season trials in Tanzania and
yield and leaf length and width increased at later harvests Taiwan, where high yielding accessions showed lower oxalate
(Table 2). In the cool season trial, yield of the second harvest accumulation. No significant association between oxalate and
was strongly correlated to leaf length and width; in subsequent mineral accumulation was found except for the cool season
harvests, this correlation decreased. In Harvests 3 and 4, only trial in Taiwan, in which oxalate levels were correlated with
leaf length remained significantly correlated with yield, indi- Ca and Zn levels. Association between Ca and Fe contents
cating other factors than leaf morphology contributed yield were significant across seasons and environments (Table 4).
in later harvests of the cool season trial. There was no sig- Figure 3 illustrates the differences in yield and concentrations
nificant association detectable between insect pest or disease of oxalate, Ca, Fe, and Zn among the trials in the cool and
incidence with yield. In the trials in Taiwan, the available data warm seasons in Taiwan and Tanzania. Overall, the analysis
on leaf size were not sufficient to allow determination of cor- suggests that in terms of yield, the accessions behave quite
relations with yields. similarly in the warm season in Tanzania and the cool season
in Taiwan (r = .62, p < .01, Figure 3, Supplemental Table S3).
The response of accessions in terms of oxalate and Fe accu-
3.3 Evaluation of dry matter, oxalate, and mulation is similar across seasons and environments except
mineral content in the CC for Tanzania in the warm season, and for Ca across the cool
seasons at both locations.
Dry matter, as well as mineral concentrations (Ca, Fe, and
Zn) and oxalate were measured in edible plant parts harvested
during the yield trials in Tanzania and Taiwan. The variation 3.4 Variation in drought tolerance
found in the CC for dry matter content was 1.5-fold at each
site and season, and for oxalate and Ca about two-fold. For A pretest assessing wilting of CC accessions upon with-
Fe, the variation in the germplasm set was the highest in the holding irrigation revealed variation among the germplasm
warm season in Tanzania and the lowest in the warm season in (Figure 4). No significant correlation between morphological
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1180 Crop Science SCHAFLEITNER ET AL.

T A B L E 3 Minimum (Min.), average, maximum (Max.), and coefficient of variation (CV) values for yield, dry matter (DM), oxalate (Ox), and
calcium (Ca), iron (Fe), and zinc (Zn) in edible plant parts harvested from the amaranth core collection grown in Arusha, Tanzania, and Shanhua,
Taiwan

Variable Yield DM Ox Ca Fe Zn Yield DM Ox Ca Fe Zn

g plot−1 % mg 100 g−1 fresh weight g plot−1 % mg 100 g−1 fresh weight
Shanhua, Taiwan, warm season Shanhua, Taiwan, cool season
Min. 130 10.1 180.8 226.7 2.8 0.4 1,670 9.8 250.4 362.6 2.1 0.3
Max. 7,966 15.0 382.1 653.8 5.8 1.4 17,820 16.0 618.4 773.3 7.7 1.9
Average 3,247 12.8 261.6 402.9 4.0 0.7 11,965 13.1 375.2 529.2 4.0 0.6
SD 1,816 1.0 45.1 89.4 0.7 0.2 3,323 1.4 66.2 93.4 1.0 0.3
CV 55.9 7.9 17.2 22.2 17.6 27.5 27.8 10.6 17.6 17.6 25.8 44.2
Arusha, Tanzania, warm season Arusha, Tanzania, cool season
Min. 3,213 10.8 272.7 356.7 5.9 0.4 530 11.7 300.5 364.5 5.9 0.3
Max. 14,873 15.9 515.0 766.0 26.4 1.6 11,879 18.9 555.2 847.3 15.2 2.4
Average 10,956 13.5 386.4 477.2 9.8 0.6 5,648 13.8 401.5 580.2 9.0 0.6
SD 2,056 0.9 56.1 75.9 3.6 0.2 2,601 1.4 56.9 111.9 2.0 0.4
CV 18.8 6.9 14.5 15.9 37.1 30.5 46.0 10.0 14.2 19.3 22.5 61.5

T A B L E 4 Correlations (r) between yield, dry matter (DM), and nutritional parameters such as oxalate (Ox), calcium (Ca), iron (Fe), and zinc
(Zn) during the warm and the cool season in Tanzania and Taiwan

Variable DM Ox Ca Fe Zn DM Ox Ca Fe Zn
Tanzania warm season (n = 78) Tanzania cool season (n = 78)
Yield .11 .01 −.13 −.19 .01 −.23 −.36** −.60** −.21 −.08
DM .10 .56** −.03 .34** .31** .57** .45** .20
Ox .05 .06 .01 .18 −.19 .22
Ca .25* .07 .53** .18
Fe .17 .13
Taiwan warm season (n = 69) Taiwan cool season (n = 69)
Yield −.03 .14 .37** .04 .49 −.12 −.34** −.22 .08 −.33**
DM −.12 .64**
.59 **
.33 **
.16 .65**
.35 **
.00
Ox −.05 .02 −.03 .31* −.09 .24*
Ca .69** .61** .25* .28*
Fe .32** −.20

*Significant at the .05 probability level.

**Significant at the .01 probability level.

features such as plant height, leaf length or width, and wilting showed relatively low wilting under drought stress. Likewise,
under drought could be detected, indicating traits other than the local checks AM#53 and AM#74 were relatively tolerant
leaf morphology contributed to reduced wilting. However, to drought.
wilting scores on Days 6 and 9 were positively correlated with
yield (r = .31, p < .05), indicating high yielding accessions
wilted earlier than low yielding accessions. On Day 12 after 4 DISCUSSION
withholding irrigation, no association between wilting under
drought and yield under normal conditions was detected. The Traditional vegetables such as amaranth have the potential
most drought susceptible accessions belonged to A. dubius, to contribute essential nutrients to diets and generate income
and the most tolerant to A. caudatus and A. cruentus. Wilt- for farmers (Mwadzingeni et al., 2021), but productivity is
ing scores of the amaranth cultivars ranged around the wilting constrained due to a lack of access to quality seed (Ndinya
score average of the germplasm panel, except Madiira 2 which et al., 2020). A few improved amaranth cultivars were
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHAFLEITNER ET AL. Crop Science 1181

F I G U R E 3 Principal component analysis of yield (a) and concentrations of (b) oxalate. Calcium (Ca; c), iron (Fe; d), and zinc (Zn; e) in the
core collection (CC) accessions across environments World Vegetable Center (WorldVeg) headquarters (HQ) and WorldVeg Tanzania (Arusha). The
CC accessions are numbered according to Supplemental Table S1

introduced and were well adopted by farmers, but more agricultural sustainability (Sequeros et al., 2021). A crucial
cultivars, especially those with tolerance to biotic stress, requirement for developing improved cultivars is access to
are required (Ochieng et al., 2019). Developing improved genetic variation of traits for breeding. Genebanks worldwide
cultivars with farmer-requested traits is important to meet the conserve the diversity of crops and crop wild relatives
contemporary challenges of improving human nutrition and and serve as a source of germplasm and traits for crop
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1182 Crop Science SCHAFLEITNER ET AL.

F I G U R E 4 Drought tolerance pretest. Wilting was scored from 0 (no wilting) to 6 (plant was dried out). CAU, A. caudatus; CRU, A. cruentus;
DUB, A. dubius; HYP, A. hypochondriacus; ND, not determined; D6, D9, and D12: Day 6, 9, 12 after withholding irrigation; AV, average of D6, D9,
and D12

improvement (Ebert, 2020). Screening large collections for was preferred (Mncwango et al., 2021), but more organoleptic
traits of interest can be laborious, especially when multiloca- analyses are required to identify preferred tastes of vegetable
tion evaluation of the material for multiple traits is involved. amaranths across regions and consumer groups.
Core collections represent a large portion of the diversity Amaranth is rich in minerals like macro- and microele-
present in large collections. These relatively small germplasm ments, such as nitrogen (N), P, K, Ca, Mg, Fe, manganese
sets can be screened for traits of interest at lower cost (van (Mn), copper (Cu), sodium (Na), and Zn (Chakrabarty et al.,
Hintum et al., 2000). Access to diversity at lower cost is 2018; Sarkar et al., 2016); high quality protein with essen-
particularly important for small breeding programs with tial amino acids (Sakar et al., 2014); and vitamins (Sarkar
limited budgets. A CC was developed for amaranth, a crop et al., 2015). One breeding target could be to improve the
species with extraordinary potential to diversify agricultural nutrient content of the grain and the leaves even further. How-
production and contribute a highly nutritious grain and a ever, the grain also can contain variable amounts of antinutri-
nutritious leafy vegetable to human diets. Although the size ents such as phytate, saponins, tannins, oxalates, and small
of the CC was minimized for easy screening, the population amounts of protease inhibitors, and the leaves contain oxalate
genetic parameters and the conservation of a large part of the and small amounts of saponins and nitrates (Aderibigbe et al.,
trait variation present in the WC indicate the CC represents a 2020). The effects of oxalates and phytates in inhibiting nutri-
large portion of the diversity of the WorldVeg germplasm col- ent uptake have raised concerns when promoting consumption
lection of the four species mostly used for breeding amaranth of leafy vegetables. Processing can lower antinutrient contents
cultivars for Africa: A. cruentus, A. hypochondriacus, A. of amaranth (Babatunde & Gbadamosi, 2017; Njoki, 2015)
caudatus, and A. dubius. For all analyzed traits except for leaf and cooking methods have been developed to lower the effect
size, the CC provides variation beyond the one found in cur- of antinutrients and enhance nutrient uptake (Nyonje et al.,
rent cultivars. Current breeding aims for vegetable amaranth 2021). The antinutrient concentration also could be lowered
with high leaf yields, delayed flowering, tolerance to drought by breeding, but this could lead to a reduction of antimicrobial
and heat stress, good taste, and high nutrient content (Adeniji or insecticidal activities in the plant (Soetan, 2008) and result
& Aloyce, 2013; Joshi et al., 2018). Variation has been found in cultivars more susceptible to insect pests and diseases.
in the CC for most of these traits. Accessions with higher Oxalate and Ca concentrations in amaranth were correlated
yields, tolerance to drought, and higher nutrient content than in the cool season in Taiwan. A correlation between Ca and
current cultivars have been identified, suggesting that further oxalate would not be surprising, as a part of the Ca in the
improvement of cultivars using the CC as a source for new cell is present in the form of Ca oxalate crystals. Oxalate
traits should be possible. However, taste was not evaluated has been reported to be synthesized from glycolate, and to a
in the CC. Previous research indicated A. dubius cultivars lesser extent from L-ascorbic acid in spinach (Spinacia oler-
are often preferred for the taste of their leaves (Ochieng acea) leaves (Fujii et al., 1993). Calcium oxalate crystal for-
et al., 2019), whereas other research found no association mation in plants has been attributed various roles, includ-
between specific species and sensory properties, as cultivars ing tissue Ca regulation, defense against herbivory, and metal
of the same species had partly showed contradictory sensory detoxification (Franceschi & Loewus, 1995; He et al., 2013).
properties (Hiscock et al., 2018). In South Africa, mild taste Tooulakou et al. (2016), Nakata (2012), and Webb (1999;
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHAFLEITNER ET AL. Crop Science 1183

2019) hypothesized Ca oxalate crystals function as carbon days were without precipitation, except for the last 2 wk,
stores that can be activated under drought conditions, when where a few rainy days were recorded. The relatively warm
stomata are closed. Tooulakou et al. (2016) demonstrated that weather combined with light rain in the warm season appar-
partial stomatal closure during the day is accompanied by a ently generated overall better conditions for amaranth growth
gradual decrease of the Ca oxalate crystal volume which is than the dry and cooler climate of the cool season trial in Tan-
recovered during the night, confirming that Ca oxalate crys- zania. The insect pest and disease incidence in the field dur-
tals are indeed a dynamic system. The reduction of the Ca ing the cool season was very low and did not significantly
oxalate crystal amounts during “alarm photosynthesis,” when damage the plants. The greatest differences in yield between
stomata are closed, has been shown in amaranth (A. hybridus), the warm and the cool season appeared in the later harvests,
but also in C3 plants and in the CAM plant Portulacaria afra, and many accessions did not yield at the fourth and fifth
suggesting Ca oxalate may be used as a carbon store in many harvest.
plant families. Amaranth, as a C4 species, reduces photores- The demand for amaranth is rising, generating opportuni-
piration by taking up CO2 in the darkness when transpiration ties for farmers to market additional grain and vegetable ama-
rates are low, and stores it as malate which can be transported ranth. For example, in Kenya maize (Zea mays) flour is for-
to the bundle sheet cells during the day where photosynthe- tified with milled amaranth seed to increase protein, Fe, and
sis is taking place. This minimizes interference from oxy- Zn content (Kamotho et al., 2017) which increases demand
gen, thereby avoiding carbon loss through photorespiration. for amaranth grain. Amaranth grain is also promoted as func-
Accumulation of Ca oxalate crystals may be an additional tional food (Martinez-Lopez et al., 2020), used as a no-gluten
mechanism to allow for photosynthesis while stomata remain alternative to cereals in bakeries (Gebreil et al., 2020), for bev-
closed, resulting in a better adaptation of the plant to stress- erages (Manassero et al., 2020) including beer (Yang & Gao,
ful conditions. Taking into consideration the relatively low 2020), and to extract a high quality oil for cosmetics (Huang
variation of oxalate content among amaranth genotypes and et al., 2009) and food applications (Becker, 2018), to mention
the putative importance of oxalate in stress tolerance, it may just a few current uses. Vegetable amaranth is increasingly
be favorable to consider preparation methods that minimize popular in Asia and Africa, but vegetable amaranth marketing
the antinutritive effects of Ca oxalate rather than select culti- faces limitations in Africa that restrict the economic benefits
vars with low oxalate concentrations. Interestingly, on average for farmers (Dizyee et al., 2020). Dual-use cultivars for har-
the lowest oxalate levels were found during a dry period of vesting leaves multiple times during the growing season and
the warm season trial at WorldVeg HQ, where transpirational for a final grain harvest could contribute to satisfy the rising
demand likely was the highest compared with the other tri- demand for amaranth grain and leaf while allowing farmers
als. Additional research is required to verify diurnal changes to generate income with two products instead of one (Dinssa
of oxalate concentration in vegetable amaranth, which could et al., 2018). Through good choice of harvesting intensity,
lead to the introduction of harvesting methods to minimize the cultivar, and environment, the grain yield penalty due to
oxalate content in the produce. leaf harvest can be minimized (Hoidal et al., 2019). Ama-
The agronomic evaluation in Tanzania and in Taiwan over ranth cultivars varied in grain yield in response to defoliation.
two seasons showed the germplasm panel performs the best at Entries PARIS (AVAM1606) and BRESIL (AVAM1607) pro-
an average temperature around 21 ˚C with minimum temper- duced high grain yields only if plants were not defoliated;
atures around 15˚C and a maximum around 30 ˚C. Reduction entries IP-5-Sel, SIMON’S FARM (AVAM1601) and Madiira
of the temperature of on average 3 ˚C leads to yield losses of 2 yielded ample leaves but relatively little grain; and entries
in average 50% whereas a rise of the average temperatures by such as AH-TL-Sel, AH-NL (AVAM1603), TZSMN102-
6 ˚C and of the minimal temperatures by 7 to 8 ˚C causes yield Sel, and ‘Mchicha’ yielded moderate amounts of grain and
losses of more than 70% on average. The warm season in Tan- leaf (Dinssa et al., 2018). We suggest using the CC to
zania and the cool season in Taiwan were the most appropri- screen for genotypes with vigorous early leaf growth, late
ate growing seasons for the amaranth germplasm panel. Some flowering, and strong seed set as candidates for dual use
accessions like VI044475 and VI050128 produce acceptable cultivars.
yield of around 50% of their yield potential under optimal As C4 plants, amaranth can efficiently tolerate abiotic
conditions, also under hot conditions in Taiwan, whereas in stresses such as heat (Hwang et al., 2018) and drought (Sarkar
Tanzania selection IP-2 AVAM1602 (VI061472) and culti- & Oba, 2018a). In addition, amaranth is relatively tolerant
var Madiira 2 (VI060470) had greater or nearly as much veg- to salinity stress (Sarkar & Oba, 2018b). Both drought and
etable yield during the cool season as in the warm season. salinity stress cause oxidative damage (Sarkar & Oba, 2018c),
No reliable precipitation data were available for the field tri- and the presence of both antioxidant enzymes and nonen-
als in Tanzania, but data from a weather station in the region zymatic antioxidants may contribute to the detoxification of
indicates that on most days of the warm season trial was at reactive oxygen species (Sarkar & Oba, 2020). Moreover, the
least a little rain, whereas during the cool season trial, most high drought and salt tolerance of amaranth is associated with
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1184 Crop Science SCHAFLEITNER ET AL.

a low basal stomatal conductance due to a low number of 5 CONCLUSION

stomata and low degree of stomata aperture, supported by the
C4 metabolism, which contributes to avoid water loss under A CC representing the genetic and phenotypic diversity of
osmotic stress (Estrada et al., 2021). The relative osmotic amaranth species with great importance in Africa held by
stress tolerance makes amaranth an option for horticultural the WorldVeg genebank was produced with the intention to
production in dry environments, and makes it suitable for use facilitate access to the biodiversity of this crop for breeding-
in various other environments and production systems, includ- improved vegetable and dual use amaranth cultivars. Analysis
ing urban agriculture. Nevertheless, the number of cultivars of available descriptor data of the CC and WC corroborated
for grain and vegetable production available for farmers is lim- that the CC represents a large part of the diversity of the WC.
ited, which warrants investment in breeding. The CC was completed by including cultivars and selected
Multilocation trials demonstrated that amaranth leaf yields lines. Evaluation of the core collection over two seasons in two
are strongly affected by the environment. Both trial locations environments identified accessions and cultivars in the CC
were located in the tropics with small day length variation with high and stable yields. The most stressful environment
over the year, suggesting variation in day length had only a in which the CC was tested in this study was the warm sea-
minor effect on yield, if at all. But the yield comparisons in son in Taiwan, where hot temperatures limit crop productivity;
the present study between sites in Tanzania and Taiwan were nevertheless, several genebank accessions were identified that
not only influenced by climatic conditions. The cultivation yielded under these stressful conditions. Likewise, amaranth
methods were also different. Taiwan farmers prefer harvest- lines yielding relatively well in both the cool and warm season
ing from younger plants and harvest young shoots in shorter in Tanzania were identified. Yield and content of Fe and Ca
intervals, whereas farmers in Tanzania harvest leaves at about were not associated, suggesting improvement of both traits is
2-wk intervals. Therefore, the evaluation of the accessions possible without yield or nutrient penalty. The CC, but also
not only accounted for different environments and seasons, all accessions of the WC, are made available by WorldVeg
but also for different local cultivation practices. The World- to the public under a standard material transfer agreement for
Veg developed improved cultivar Madiira 2 stood out in the breeding, research and training for food and agriculture pur-
warm and cool season trials in Tanzania and in the cool sea- poses (https://avrdc.org/our-work/managing-germplasm/).
son trial for relatively high yield, ranking near the top among
all accessions. Wilting under drought was delayed in this cul- AC K N OW L E D G M E N T S
tivar, but the mineral content of this line was not above aver- We thank the laboratory, field, and greenhouse teams at
age. As yield and Fe and Ca content are not associated during WorldVeg HQ at Shanhua, Taiwan and at the World-
the seasons when amaranth provides high yields, an increase Veg Center Eastern and Southern Africa, Arusha, Tan-
of Ca and Fe should be possible without negatively affecting zania for their great support. This work received finan-
yield. cial support from the German Federal Ministry for Eco-
The average Fe amounts of 100-g amaranth leaves har- nomic Cooperation and Development (BMZ) commissioned
vested in the trials in Tanzania exceed the recommended daily by the Deutsche Gesellschaft für Internationale Zusamme-
Fe intake for women at reproductive age of 14.8 mg (NHS, narbeit (GIZ) through the Fund International Agricultural
2021). In Taiwan, the Fe levels in amaranth leaves were half as Research (FIA), Grant No. 81219429. Additional support
high, probably due to differences in Fe concentration and/or was provided by long-term strategic donors to the World
availability in the soil. For Ca, 100-g fresh amaranth leaves Vegetable Center: Taiwan, UK aid from the UK govern-
contain on average half of the daily recommended dose of ment, United States Agency for International Development
800 mg. The content of Zn in amaranth leaves on average of (USAID), Australian Centre for International Agricultural
0.6 mg per 100 g fresh weight is too low to provide a sig- Research (ACIAR), Germany, Thailand, Philippines, Korea,
nificant contribution to the 9.5 mg required by humans per and Japan.
day. The proportion of the Fe and Ca in amaranth leaves,
which is actually bioavailable for humans, depends on mul- AU T H O R C O N T R I B U T I O N S
tiple factors including the presence of antinutrients and the Roland Schafleitner: Conceptualization; Data curation;
preparation method (Yang & Tsou, 2006). High contribution Funding acquisition; Investigation; Methodology; Project
of environmental conditions to mineral content suggests that administration; Resources; Supervision; Validation; Writing
breeding for high mineral content is likely less effective. Vari- – original draft. Ya-ping Lin: Data curation; Investigation;
able micronutrient bioavailability likely can be addressed by Methodology; Software; Validation; Visualization. Sognigbé
improved preparation methods (Nyonje et al., 2021). Over- N’Danikou: Data curation, Investigation. Fekadu Fufa Dinssa:
all, the high mineral content, especially for Fe and Ca, makes Conceptualization; Data curation; Investigation; Methodol-
amaranth leaves particularly useful for populations affected ogy. Richard Finkers: Investigation; Software. Ruth Minja:
by micronutrient deficiencies. Investigation. Mary Abukutsa-Onyango: Investigation.
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHAFLEITNER ET AL. Crop Science 1185

Winnie Nyonje: Investigation. Chen-yu Lin: Investigation. chains for livelihood enhancement in East Africa: A systems mod-
Tien-hor Wu: Data curation; Investigation. Jeremiah Phanuel elling approach. African Journal of Agricultural and Resource Eco-
Sigalla: Investigation. Maarten van Zonneveld: Investigation. nomics, 15, 81–94.
Early, D. (1992). The renaissance of amaranth. In N. Foster & L. S.
Yun-yin Hsiao: Investigation. Sanjeet Kumar: Investigation.
Cordell (Eds.), Chiles to chocolates: Foods the Americas gave the
Wan-jen Wu: Investigation. Hsin-I Wang: Investigation. Shou
world (pp. 15–33). University of Arizona Press.
Lin: Investigation. Ray-yu Yang: Investigation. Ebert, A. W. (2020). The role of vegetable genetic resources in nutrition
security and vegetable breeding. Plants, 9(6), 736. https://doi.org/10.
CONFLICT OF INTEREST 3390/plants9060736
The authors report no conflicts of interest. Estrada, Y., Fernández-Ojeda, A., Morales, B., Egea-Fernández, J. M.,
Flores, F. B., Bolarín, M. C., & Egea, I. (2021). Unraveling the strate-
ORCID gies used by the underexploited amaranth species to confront salt
stress: Similarities and differences with quinoa species. Frontiers in
Roland Schafleitner https://orcid.org/0000-0003-3637-
Plant Science, 12.
7603
Franceschi, V. R., & Loewus, F. A. (1995). Oxalate biosynthesis and
Ya-ping Lin https://orcid.org/0000-0002-9575-2007 function in plants and fungi. In S. R. Khan (Ed.), Calcium oxalate
Richard Finkers https://orcid.org/0000-0002-4368-8058 in biological systems (pp. 113–130). CRC Press.
Fujii, N., Watanabe, M., Watanabe, Y., & Shimada, N. (1993). Kate
REFERENCES of oxalate biosynthesis from glycolate and ascorbic acid in spinach
Adeniji, O. T., & Aloyce, A. (2013). Farmers’ participatory identification leaves. Soil Science and Plant Nutrition, 39(4), 627–634. https://doi.
of horticultural traits: Developing breeding objectives for vegetable org/10.1080/00380768.1993.10419179
amaranth in Tanzania. Journal of Crop Improvement, 27(3), 309–318. Gebreil, S. Y., Ali, M. I., & Mousa, E. A. (2020). Utilization of ama-
Aderibigbe, O. R., Ezekiel, O. O., Owolade, S. O., Korese, J. K., Sturm, ranth flour in preparation of high nutritional value bakery products.
B., & Hensel, O. (2020). Exploring the potentials of underutilized Food and Nutrition Sciences, 10(05), 336. https://doi.org/10.4236/
grain amaranth (Amaranthus spp.) along the value chain for food and fns.2020.105025
nutrition security: A review. Critical Reviews in Food Science and Genesys. (2021). Genesys. https://www.genesys-pgr.org/
Nutrition, 5, 1–4. https://doi.org/10.1080/10408398.2020.1825323 Glaubitz, J. C., Casstevens, T. M., Lu, F., Harriman, J., Elshire, R. J., Sun,
Babatunde, T. O., & Gbadamosi, S. O. (2017). Effect of different treat- Q., & Buckler, E. S. (2014). TASSEL-GBS: A high capacity geno-
ments on in vitro protein digestibility, antinutrients, antioxidant prop- typing by sequencing analysis pipeline. PLOS ONE, 9(2), e90346.
erties and mineral composition of Amaranthus viridis seed. Cogent https://doi.org/10.1371/journal.pone.0090346
Food and Agriculture, 3(1), 1–14. https://doi.org/10.1080/23311932. He, H., Veneklaas, E. J., Kuo, J., & Lambers, H. (2013). Physiological
2017.1296402 and ecological significance of biomineralization in plants. Trends in
Becker, R. (2018). Amaranth oil: Composition, processing, and nutri- Plant Science, 19, 166–74. https://doi.org/10.1016/j.tplants.2013.11.
tional qualities. In O. Paredes-Lopez (Ed.), Amaranth biology, chem- 002
istry, and technology (pp. 133–141). CRC Press. Hiscock, L., Bothma, C., Hugo, A., Van Biljon, A., & Jansen Van
Bonasora, M. G., Poggio, L., & Greizerstein, E. J. (2013). Cyto- Rensburg, W. S. (2018). Overall liking and sensory profiling of boiled
genetic studies in four cultivated Amaranthus (Amaranthaceae) Amaranthus leaves using the Check-all-that-apply question. CyTA-
species. Comparative Cytogenetics, 7(1), 53. https://doi.org/10.3897/ Journal of Food, 16(1), 822–830.
CompCytogen.v7i1.4276 Hoidal, N., Díaz Gallardo, M., Jacobsen, S. E., & Alandia, G. (2019).
Bressani, R., De Martell, E. C. M., & De Godinez, C. M. (1993). Amaranth as a dual-use crop for leafy greens and seeds: Stable
Protein quality evaluation of amaranth in adult humans. Plant responses to leaf harvest across genotypes and environments. Fron-
Foods for Human Nutrition, 43(2), 123–143. https://doi.org/10.1007/ tiers in Plant Science, 10, 817. https://doi.org/10.3389/fpls.2019.
BF01087917 00817
Chakrabarty, T., Sarker, U., Hasan, M., & Rahman, M. M. (2018). Vari- Hu, J., Zhu, J., & Xu, H. M. (2000). Methods of constructing core
ability in mineral compositions, yield and yield contributing traits collections by stepwise clustering with three sampling strategies
of stem amaranth (Amaranthus lividus). Genetika, 50(3), 995–1010. based on the genotypic values of crops. Theoretical and Applied
https://doi.org/10.2298/GENSR1803995C Genetics, 101(1–2), 264–268. https://doi.org/10.1007/s00122005
Das, S. (2016). Breeding of amaranths. In S. Das (Ed.), Amaranthus: A 1478
promising crop of future (pp. 107–145). Springer. Huang, Z. R., Lin, Y. K., & Fang, J. Y. (2009). Biological and pharma-
Dinssa, F. F., Yang, R.-Y., Ledesma, D. R., Mbwambo, O., & Hanson, cological activities of squalene and related compounds: Potential uses
P. (2018). Effects of leaf harvest on grain yield and nutrient content in cosmetic dermatology. Molecules, 14, 540.
of diverse amaranth entries. Scientia Horticulturae, 236, 146–157. Hwang, S. G., Chao, H. C., & Lin, H. L. (2018). Differential responses of
https://doi.org/10.1016/j.scienta.2018.03.028 pak choi and edible amaranth to an elevated temperature. Hortscience,
Dinssa, F., Nyabinda, N., Byrnes, D. R., Ndinya, C., Merchant, E. V., 53(2), 195–199.
& Simon, J. E. (2020). Performances of vegetable amaranth entries in Joshi, D. C., Sood, S., Hosahatti, R., Kant, L., Pattanayak, A., Kumar,
yield and nutrient contents in Tanzania and Kenya, and variety release. A., Yadav, D., & Stetter, M. G. (2018). From zero to hero: The
Journal of Medicinally Active Plants, 9(3), 181–194. past, present and future of grain amaranth breeding. Theoretical
Dizyee, K., Baker, D., Herrero, M., Burrow, H., McMillan, L., Sila, and Applied Genetics, 131(9), 1807–1823. https://doi.org/10.1007/
D. N., & Rich, K. M. (2020). The promotion of amaranth value s00122-018-3138-y
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1186 Crop Science SCHAFLEITNER ET AL.

Kachiguma, N. A., Mwase, W., Maliro, M., & Damaliphetsa, A. (2015). Nyonje, W. A., Yang, R. -Y., Wu, W. -J., Makokha, A. O., Owino, W. O.,
Chemical and mineral composition of amaranth (Amaranthus L.) & Abukutsa-Onyango, M. O. (2021). Enhancing the nutritional qual-
species collected from central Malawi. Journal of Food Research, ity of vegetable amaranth through specific food preparation methods.
4(4). https://doi.org/10.5539/jfr.v4n4p92 Journal of Food Research, 10(4), 42–55. https://doi.org/10.5539/jfr.
Kamotho, S. N., Kyallo, F. M., & Sila, D. N. (2017). Biofortification of v10n4p42
maize flour with grain amaranth for improved nutrition. African Jour- Ochieng, J., Schreinemachers, P., Ogada, M., Dinssa, F. F., Barnos, W.,
nal of Food, Agriculture, Nutrition and Development, 17(4), 12574– & Mndiga, H. (2019). Adoption of improved amaranth varieties and
12588. https://doi.org/10.18697/ajfand.80.15945 good agricultural practices in East Africa. Land Use Policy, 83, 187–
Koo, D. H., Molin, W. T., Saski, C. A., Jiang, J., Putta, K., Jugulam, 194. https://doi.org/10.1016/j.landusepol.2019.02.002
M., Friebe, B., & Gill, B. S. (2018). Extrachromosomal circular Odong, T. L., Jansen, J., Van Eeuwijk, F. A., & van Hintum, T. J. (2013).
DNA-based amplification and transmission of herbicide resistance Quality of core collections for effective utilisation of genetic resources
in crop weed Amaranthus palmeri. Proceedings of the National review, discussion and interpretation. Theoretical and Applied Genet-
Academy of Sciences, 115(13), 3332–3337. https://doi.org/10.1073/ ics, 126(2), 289–305.
pnas.1719354115 Oke, O. L. (1980). Amaranth in Nigeria. In Proceedings of the Second
Lin, Y. -P., Wu, T. -H., Chan, Y. -K., van Zonneveld, M., & Schafleitner, Amaranth Conference (pp. 22–30). Rodale Press.
R. (2021). Interspecies GWAS reveals genetic divergence of days Pichop, G. N., Abukutsa-Onyango, M., Noorani, A., & Nono-Womdim,
to flowering between grain- and leaf-purpose Amaranthus species. R. (2014). Importance of indigenous food crops in tropical Africa:
bioRxiv. https://doi.org/10.1101/2021.10.05.463269 Case study. In XXIX International Horticultural Congress on Hor-
Manassero, C. A., Añón, M. C., & Speroni, F. (2020). Development ticulture: Sustaining Lives, Livelihoods and Landscapes (IHC2014)
of a high protein beverage based on amaranth. Plant Foods for (pp. 315–322). International Society of Horticultural Sciences.
Human Nutrition, 75(4), 599–607. https://doi.org/10.1007/s11130- Sarker, U., & Oba, S. (2018a). Drought stress enhances nutritional
020-00853-9 and bioactive compounds, phenolic acids and antioxidant capac-
Martinez-Lopez, A., Millan-Linares, M. C., Rodriguez-Martin, N. M., ity of Amaranthus leafy vegetable. BMC Plant Biology, 18, 258.
Millan, F., & Montserrat-de la Paz, S. (2020). Nutraceutical value of https://doi.org/10.1186/s12870-018-1484-1
kiwicha (Amaranthus caudatus L.). Journal of Functional Foods, 65, Sarker, U., & Oba, S. (2018b). Augmentation of leaf color parameters,
103735. https://doi.org/10.1016/j.jff.2019.103735 pigments, vitamins, phenolic acids, flavonoids and antioxidant activ-
Mncwango, N. C., Van Jaarsveld, C. M., Ntuli, N. G., & Mavengahama, ity in selected Amaranthus tricolor under salinity stress. Scientific
S. (2021). Participatory selection of Amaranthus genotypes in the Reports, 8, 12349. https://doi.org/10.1038/s41598-018-30897-6
KwaMbonambi area, KwaZulu-Natal, South Africa. Sustainability, Sarker, U., & Oba, S. (2018c). Drought stress effects on growth, ROS
13, 21, 11962. markers, compatible solutes, phenolics, flavonoids, and antioxidant
Mota, C., Santos, M., Mauro, R., Samman, N., Matos, A. S., Torres, D., activity in Amaranthus tricolor. Applied Biochemistry and Biotech-
& Castanheira, I. (2016). Protein content and amino acids profile of nology, 186, 999–1016. https://doi.org/10.1007/s12010-018-2784-5
pseudocereals. Food Chemistry, 193, 55–61. https://doi.org/10.1016/ Sarker, U., Islam, M. T., Rabbani, M. G., & Oba, S. (2014). Genotypic
j.foodchem.2014.11.043 variability for nutrient, antioxidant, yield and yield contributing traits
Mwadzingeni, L., Afari-Sefa, V., Shimelis, H., N’Danikou, S., Figlan, in vegetable amaranth. Journal of Food Agriculture and Environment,
S., Depenbusch, L., Shayanowako, A. I., Chagomoka, T., Mushayi, 12, 168–74. https://doi.org/10.1234/4.2014.5378
M., Schreinemachers, P., & Derera, J. (2021). Unpacking the value Sarker, U., Islam, M. T., Rabbani, M. G., & Oba, S. (2016). Genetic
of traditional African vegetables for food and nutrition security. Food variation and interrelationships among antioxidant, quality, and agro-
Security, 16, 1–2. https://doi.org/10.1007/s12571-021-01159-7 nomic traits in vegetable amaranth. Turkish Journal of Agriculture and
Nakata, P. A. (2012). Plant calcium oxalate crystal formation, function, Forestry, 40(4), 526–535. https://doi.org/10.3906/tar-1405-83
and its impact on human health. Frontiers in Biology, 7, 254–266. Sarker, U., Islam, T., Rabbani, G., & Oba, S. (2015). Genotype vari-
https://doi.org/10.1007/s11515-012-1224-0 ability in composition of antioxidant vitamins and minerals in veg-
Nampeera, E. L., Nonnecke, G. R., Blodgett, S. L., Tusiime, S. M., etable amaranth. Genetika, 47(1), 85–96. https://doi.org/10.2298/
Masinde, D. M., Wesonga, J. M., Murungi, L. K., Baidu-Forson, J. J., GENSR1501085S
& Abukutsa-Onyango, M. O. (2019). Farmers’ knowledge and prac- Sarker, U., Oba, S., & Daramy, M. A. (2020). Nutrients, minerals, antiox-
tices in the management of insect pests of leafy amaranth in Kenya. idant pigments and phytochemicals, and antioxidant capacity of the
Journal of Integrated Pest Management, 10(1), 31. leaves of stem amaranth. Scientific reports, 10(1), 1–9.
Ndinya, C., Dinssa, F., Simon, J. E., Nyabinda, N., Maiyo, N., Weller, S., Sauer, J. D. (1967). The grain amaranths and their relatives: A revised
Odendo, M., Onyango, E., Mwangi, M., & Makete, N. (2020). Assess- taxonomic and geographic survey. Annals of the Missouri Botanical
ment of seed quality of selected African leafy vegetables produced in Garden, 54, 102–137. https://doi.org/10.2307/2394998
western Kenya using informal and semi-formal seed systems. Journal Sequeros, T., Ochieng, J., Schreinemachers, P., Binagwa, P. H., Huelgas,
of Medicinally Active Plants, 9(4), 269–280. Z. M., Hapsari, R. T., Juma, M. O., Kangile, J. R., Karimi, R.,
NHS. (2021). Iron: Vitamins and minerals. https://www.nhs.uk/ Khaririyatun, N., Mbeyagala, E. K., Mvungi, H., Nair, R. M., Sanya,
conditions/vitamins-and-minerals/iron/#:∼:text=How%20much% L. N., Nguyen, T. T. L., Phommalath, S., Pinn, T., Simfukwe, E.,
20iron%20do%20I,women%20aged%2019%20to%2050) & Suebpongsang, P. (2021). Mungbean in Southeast Asia and East
Njoki, J. W. (2015). Impact of different processing techniques on nutri- Africa: Varieties, practices and constraints. Agriculture & Food Secu-
ents and antinutrients content of grain Amaranth (Amaranthus albus) rity, 10(1), 1–13. https://doi.org/10.1186/s40066-020-00273-7
(Master’s thesis). Jomo Kenyatta University of Agriculture and Tech- Shukla, S., Bhargava, A., Chatterjee, A., & Singh, S. P. (2004). Estimates
nology. of genetic parameters to determine variability for foliage yield and
 14350653, 2022, 3, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.1002/csc2.20715 by Nat Prov Indonesia, Wiley Online Library on [16/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHAFLEITNER ET AL. Crop Science 1187

its different quantitative and qualitative traits in vegetable amaranth Science and Nutrition, 8, 1–6. https://doi.org/10.1080/10408398.
(A. tricolor). Journal of Genetics and Breeding, 58(2), 169–176. 2020.1859458
Singh, B. P., & Whitehead, W. F. (1996). Management methods for pro- Yang, R. Y., & Keding, G. B. (2009). Nutritional contributions of impor-
ducing vegetable amaranth. Progress in New Crops, 511–515. tant African indigenous vegetables. In C. M. Shackleton, M. W.
Soetan, K. O. (2008). Pharmacological and other beneficial effects of Pasquini, & A. W. Drescher (Eds.), African indigenous vegetables in
antinutritional factors in plants-A review. African Journal of Biotech- urban agriculture (pp. 137–176). Routledge.
nology, 7(25). Yeh, F. C. (1999). POPGENE (version 1.3. 1). http://www.ualberta.ca/
Tooulakou, G., Giannopoulosm, A., Nikolopoulosm, D., Bresta, P., ∼fyeh/popgene

Dotsika, E., Orkoula, M. G., Kontoyannis, C. G., Fasseas, C.,

Liakopoulos, G., Klapa, M. I., & Karabourniotis, G. (2016). Reevalu- S U P P O RT I N G I N F O R M AT I O N
ation of the plant “gemstones”: Calcium oxalate crystals sustain pho-
Additional supporting information may be found in the online
tosynthesis under drought conditions. Plant Signaling and Behavior,
11(9), 00111. https://doi.org/10.1080/15592324.2016.1215793
version of the article at the publisher’s website.
Tooulakou, G., Nikolopoulos, D., Dotsika, E., Orkoula, M. G.,
Kontoyannis, C. G., Liakopoulos, G., Klapa, M. I., & Karabourniotis,
G. (2019). Changes in size and composition of pigweed (Amaran-
thus hybridus L.) calcium oxalate crystals under CO2 starvation con- How to cite this article: Schafleitner, R., Lin, Y.-P.,
ditions. Physiologia Plantarum, 166(3), 862–872. https://doi.org/10. Dinssa, F., N’Danikou, S., Finkers, R., Minja, R.,
1111/ppl.12843 Abukutsa-Onyango, M., Nyonje, W., Lin, C.-Y., Wu,
van Hintum, T. J., Brown, A. H. D., & Spillane, C. (2000). Core collec- T.-H., Sigalla, J. P., van Zonneveld, M., Hsiao, Y.-Y.,
tions of plant genetic resources (No. 3). Bioversity International. Kumar, S., Wu, W.-J., Wang, H.-I., Lin, S., & Yang,
Webb, M. A. (1999). Cell-mediated crystallization of calcium oxalate in R.-Y. (2022). The World Vegetable Center
plants. Plant Cell, 11, 751–761. https://doi.org/10.1105/tpc.11.4.751
Amaranthus germplasm collection: Core collection
Yang, R. Y., & Tsou, S. C. (2006). Enhancing iron bioavailability of veg-
etables through proper preparation-principles and applications. Jour-
development and evaluation of agronomic and
nal of International Cooperation, 1(1), 107–119. nutritional traits. Crop Science, 62, 1173–1187.
Yang, D., & Gao, X. (2020). Progress of the use of alternatives to https://doi.org/10.1002/csc2.20715
malt in the production of gluten-free beer. Critical Reviews in Food