Genet Resour Crop Evol (2024) 71:1221–1239

https://doi.org/10.1007/s10722-023-01686-6

RESEARCH ARTICLE

Genetic diversity of kale (Brassica oleracea L. var acephala)

using agro‑morphological and simple sequence repeat (SSR)
markers
Barbara Pipan · Mohamed Neji ·
Vladimir Meglič · Lovro Sinkovič

Received: 18 May 2023 / Accepted: 12 July 2023 / Published online: 23 August 2023
© The Author(s) 2023

Abstract Kale (Brassica oleracea. var. acephala) both MFA and UPGMA clustering analysis. The SSR
is a nutrient-rich green leafy vegetable consumed markers were highly informative with 108 alleles and
as food and used in traditional medicine worldwide. polymorphic information content ranging from 0.395
An essential step in describing the available genetic to 0.856. Strong genetic diversity was detected at the
resources and ensuring their effective use in breed- accession level (H’ = 0.58) while genetic differentia-
ing programs is to characterize the genetic diversity tion was low (Fst = 0.05). Similar to UPGMA cluster-
of available germplasm. In this study, the genetic ing, Bayesian clustering suggests that the kale col-
diversity and structure of 26 kale accessions from lection can be divided into four clusters with a high
South-East Europe were examined using 26 agro- degree of admixture and no geographic grouping pat-
morphological traits collected in the field and 12 sim- tern is apparent. Overall, the study showed that the
ple sequence repeat (SSR) markers. Considerable kale collection studied represents a valuable reservoir
agro-morphological variability was found in most of genetic and agro-morphological variability that
quantitative (CV = 17.26–42.42%) and qualitative could be used for future breeding initiatives.
(H’ = 0.61–1.79) traits. Multifactorial analysis (MFA)
showed that country of origin (33.01%) and morpho- Keywords Kale · Genetic differentiation · Allelic
type (32.30%) significantly influenced kale diver- diversity · SSR markers · Plant descriptors
sification. Leaf blade shape (20.62%), leaf incision
(19.43%), anthocyanin distribution (16.43%), and
leaf colour (15.55%) were the traits that most clearly Introduction
differentiated accessions. The three common com-
mercial kale cultivars were identified as independent The diploid species (2n = 18) Brassica oleracea L.
outliers that differed from the other kale accessions in (Brassicaceae), the most diverse species in the genus
Brassica, includes several vegetable crops with a
long history of cultivation and domestication world-
Supplementary Information The online version wide, such as cauliflower, cabbage, broccoli, Brussels
contains supplementary material available at https://​doi.​
org/​10.​1007/​s10722-​023-​01686-6.
sprouts, kohlrabi, and kale (Golicz et al. 2016). In
particular, kale leaves (B. oleracea var. acephala) has
B. Pipan (*) · M. Neji · V. Meglič · L. Sinkovič attracted great interest due to its excellent nutritional
Crop Science Department, Agricultural Institute value compared to other crops. It is considered an
of Slovenia, Hacquetova ulica 17, 1000 Ljubljana,
ideal source of vitamins, essential minerals, and fatty
Slovenia
e-mail: barbara.pipan@kis.si acids (Pathirana et al. 2017). Therefore, it occupies

Vol.: (0123456789)
13
 1222 Genet Resour Crop Evol (2024) 71:1221–1239

a privileged position in the cuisine and diet of Euro- contrast, DNA-based molecular markers are stable,
pean, Asian and American populations (Gonçalves intact, abundant in the genome, and free from envi-
et al. 2012). Moreover, it is widely used in traditional ronmental influences. Therefore, they are a powerful
medicine for the treatment of rheumatism, liver dis- tool not only for comprehensive depiction the genetic
eases, eye problems, bone weakness, anaemia, and diversity patterns, but also for other advanced analy-
obesity (Gonçalves et al. 2012; Kuerban et al. 2017; ses such as gene mapping and molecular marker-
Thavarajah et al. 2016) and also as an animal feed assisted breeding (Zhu et al. 2019). Although vari-
(Cartea et al. 2003). ous molecular markers have been developed to date,
Brassica species are known as a highly diverse simple requence Repeats (SSRs) or microsatellites
group of plants due to their ability of spontaneous remain the most attractive for genetic variability
cross-pollination and gene flow among sexually com- research and plant breeding because they are codomi-
patible relatives (Meglič and Pipan 2018; Pipan et al. nantly inherited, abundant, multi-allelic and highly
2013, 2011). B. oleracea var. acephala is classified polymorphic (Riangwong et al. 2020; Rivera et al.
into different morphotypes based on morphological 2016), and are also easy to handle in the labora-
characteristics. For example, kale (B. oleracea L. var. tory or automated by capillary sequencers (Schuelke
acephala DC.) has dark green and crinkled leaves, 2000). Therefore, assessment of genetic variation
Scotch kale (B. oleracea L. var. acephala (DC.) Alef. using agro-morphological traits and SSR markers has
var. sabellica L.) is characterized by grey-green and proven to be a very powerful approach for genetic
highly crinkled and wrinkled leaves, while Marrow resources management, utilization, and conservation
stem kale (B. oleracea L. var. acephala (DC.) Alef. in many crops, including chickpea (Ghaffari et al.
var. medullosa L.) is characterized by a soft and thick 2014), tomato (Mercati et al. 2015), pepper (Rivera
stem and different leaf types. In addition to variation et al. 2016), leafy mustard (Sharma et al. 2020), and
among morphotypes, populations/cultivars within the cauliflower (Rakshita et al. 2021).
same morphotype may also exhibit wide morphologi- In the present study, a large set of quantitative
cal variation due to a long history of domestication and qualitative agro-morphological traits and SSRs
and adaptation to environmental conditions (Hahn markers were used to evaluate the genetic diversity
et al. 2022). According to Hahn et al. (2022), three and pattern of genetic structure of B. oleracea L. var.
types of kale are distinguished according to their ori- acephala collection from South-East Europe. The
gin: curly kale (Scotch type, Brassica oleracea covar. results of the study will be very useful for further
acephala var. sabellica), Italian kale (Lacinato type, use of the analysed accessions in ongoing and future
Brassica oleracea covar. acephala var. palmifolia), breeding programs.
and collard (Brassica oleracea covar. acephala var.
viridis). Indeed, in B. oleracea var. acephala, large
morphological variation could occur both within pop- Materials and methods
ulations as a result of cross-pollination and between
populations due to extensive selection by farmers Plant material
and/or adaptation to local environments (Cartea et al.
2003; Šamec et al. 2019a; Thavarajah et al. 2016). The plant material used in the study consists of
Detailed agro-morphological characterization and 26 accessions of B. oleracea var. acephala, of which
assessment of patterns of genetic variability based 23 accessions were B. oleracea L. var. acephala DC
on quantitative and qualitative traits is the first and obtained from the two national plant gene banks, i.e.
crucial step in describing available genetic resources Croatia (six accessions) and Bosnia and Herzego-
and their efficient use in breeding programs (Balkaya vina (17 accessions). The remaining three accessions
and Yanmaz 2005; Oumata et al. 2023). However, were commercial cultivars developed by three differ-
although agro-morphological traits can be controlled ent seed producers. ’Kodrolistni ohrovt’ and ’Krmni
at the genetic level, they may be nonspecific and non- ohrovt’ were two cultivars of Scotch kale and Marrow
polymorphic, and their variation could be strongly stem kale developed by Semenarna Ljubljana (Slo-
influenced by environmental factors (Choudhury venia) and Semina Royal Seeds (Slovenia), respec-
et al. 2022; Petit et al. 2020; Terlević et al. 2023). In tively, while ’Ohrovt Nero di Toscana’ was a cultivar

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1223

of Lacinato kale, B. oleracea var. palmifolia devel- [PETIOLECOLOUR] were characterised. The full
oped by Franchi Sementi (Italy). All available pass- description of all analysed traits is given in the Sup-
port data for the studied accessions are summarised in plementary Table S2. In addition to agro-morpholog-
Supplementary Table S1. ical characterization, a young healthy leaf was taken
from 2–3 plants per accession and stored at –80 °C
Experimental design and agro‑morphological for further molecular characterization. Non-healthy
characterization plants were excluded from agro-morphological and
molecular characterization. The experiment was con-
In March 2019, four individuals from each acces- ducted at the experimental fields of the Agricultural
sion were initially cultivated in seedling trays filled Institute of Slovenia in Jablje (304 m a.s.l.; 46.151°N
with fertilized peat. Two months after germination, 14.562°E).
seedlings were transplanted to open-field condi-
tions and grown to maturity. Adjacent plants were Molecular analysis
spaced 50 cm apart and a black polyethylene cover
was placed on the soil before planting. Each plant First, genomic DNA was extracted from 100 mg
was phenotypically characterised using 11 quanti- young and healthy leaves from 2 to 3 individuals per
tative and 15 qualitative non-destructive descrip- accession as described in (Pipan et al. 2017). DNA
tors for Brassica spp. The whole plant, entire leaf, from all individuals was then checked for quality
leaf blade and petiole were characterised accord- and quantity using a fluorimeter (Qubit 3.0; Ther-
ing to the European Cooperative Programme for moFisher Scientific, MA, USA) and then diluted to a
Plant Genetic Resources developed by the Com- uniform final concentration of 5.6 ng/μL and stored at
munity Plant Variety Office (CPVO 2011) and the –20 °C for further analysis.
International Board for Plant Genetic Resources DNAs from a total of 71 individuals were used
(IBPGR 1990). Quantitative traits included plant separately (2–3 individuals per accession) and then
height [PLANTHEIGHT], plant diameter [PLANT- amplified with a set of 12 genome-specific SSR mark-
DIAM], plant height/diameter ratio [PLANTRA- ers previously used to analyse the genetic diversity
TIO], leaf length [LEAFLENGTH], leaf area [LEA- of different B. oleracea crops (kale, cabbage, cauli-
FAREA], leaf blade length [BLADELENGTH], leaf flower, and Brussels sprouts) and found to be highly
blade width [BLADEWIDTH], leaf blade width/leaf polymorphic (El-Esawi et al. 2016). PCR amplifica-
length ratio [BLADERATIO], petiole length [PETI- tion was performed using fluorescently labelled uni-
OLELENGTH], petiole width [PETIOLEWIDTH], versal primers according to the protocol described
and petiole length/width ratio [PETIOLERATIO]. by (Schuelke 2000). First, for each SSR marker, the
For qualitative traits, plant shape [PLANTSHAPE] forward primer was labelled with the M13 sequence
and position of growing point relative to the top of the (5’-CAC​GAC​GTT​GTA​AAA​CGA​C-3’). The PCR
plant [PLANTPOSIT] were used to characterise plant reaction was performed in a final volume of 11 μL
shape. Leaf division (margin) [LEAFMARGIN], containing 5.6 ng genomic DNA, 1 μL 10 × PCR
leaf division (incision) [LEAFINCISION], leaf apex buffer (Biotools, Spain), 0.5 μL 50 mM M ­ gCl2
shape [LEAFAPEX], leaf tip attitude [LEAFTIP], (Biotools, Spain), 0.2 μL each 10 mM dNTPs
leaf lamina attitude [LEAFLAMINA], leaf antho- (Sigma-Aldrich, USA), 0.1 μL 10 μM forward primer
cyanin coloration [LEAFANTHO], distribution of (Sigma-Aldrich, USA), 0.25 μL 10 μM reverse primer
anthocyanin coloration on the leaf [LEAFDISTRIB], (Sigma-Aldrich, USA), 0.183 μL 10 μM 5’-fluores-
leaf colour [LEAFCOLOUR], and intensity of the cently labelled universal primer (with 6-FAM, NED
colour of the fully developed leaf [LEAFINTENS] or HEX; Omega, Slovenia), and 0.5 μL 5 U Taq DNA
were evaluated to describe leaf morphology. The polymerase (Biotools, Spain). PCR amplification was
blade shape [BLADESHAPE] and blade blistering performed in a thermal cycler (Veriti, ThermoFisher
[BLADEBLISTER] were also used to characterise Scientific) using the following primer-specific touch-
the leaf blade. Finally, the petiole by assessing petiole down profile: 94 °C for 4 min; 15 cycles at 94 °C
and/or midvein enlargement [PETIOLEMEDVEIN] for 1 min; temperature reduction from 60 (62) °C
and the colour of the petiole and/or midvein colour to 49.5 (51.5) °C at 0.7 °C per cycle for 30 s; 72 °C

Vol.: (0123456789)
13
 1224 Genet Resour Crop Evol (2024) 71:1221–1239

for 1 min; followed by 23 cycles at 94 °C for 30 s; relationship between accessions UPGMA clustering
53 °C for 30 s; 72 °C for 1 min; and final extension using the "hclust" function.
for 5 min at 72 °C.
Ultimately, an Applied Biosystems ABI3130XL SSR data analysis
genetic analyser was used to resolve the DNA frag-
ments. The allele sizes were then determined by com- The data matrix resulting from allele sizing was first
parison with an internal size standard (GeneScan-500 analysed using Power Marker version 3.25 with the
ROX; Applied Biosystems) using the GeneMapper default parameters (Liu and Muse 2005) to calcu-
6.0 software (Applied Biosystems). late several measures describing the genetic diver-
sity at the loci and accession level, including number
of different alleles (Na), effective number of alleles
Agro‑morphological data analysis (Ne), major allele frequency (MAF), and polymor-
phic information content (PIC). We also used the
The obtained data were statistically analysed with a default parameter settings in GenAlEx 6.5.2 software
series of univariate and multivariate approaches using (Peakall and Smouse 2006) to calculate the Shannon
R statistical programming environment version 3.4.4 information index (I), observed heterozygosity (Ho),
R Core Team (2021). First, the package "ggplot2" and expected heterozygosity (He), as well as the
(Villanueva and Chen 2019) was used to visualise the fixation index (F) and the number of private alleles
frequency of distribution of all agro-morphological (No. PA). The same software was also used for the
traits across accessions using histograms. To describe analysis of molecular variance (AMOVA) to estimate
patterns of the agro-morphological variation, the the proportion of variation explained within acces-
package "summarytools"(Comtois and Comtois 2016) sions, between accessions, and between geographic
was used to compute basic descriptive statistics, origins of accessions. In addition, to examine the
including maximum and minimum values (max and genetic structure of the studied accessions using the
min), standard deviation (SD), and coefficient of vari- data from the 12 SSR loci, Bayesian clustering was
ation (CV) for the quantitative traits. Analysis of vari- performed using the software STRU​ CTU​ RE 2.3.4
ance (ANOVA) was used to partition variation within (Pritchard et al. 2000). The analysis was performed
and between accessions using the "aov" function of using the Admixture model by adjusting the number
the "Stats" package (R Core Team 2021). For qualita- of populations (K) from 1 to 10 based on 10 inde-
tive traits, the "vegan" package (Oksanen et al. 2019) pendent simulations for each K. Each simulation con-
was used to compute the Shannon–Weaver diversity sisted of a burn-in period of 10.000 steps and 100.000
index (Shannon and Weaver 1949) as a measure of Markov Chain Monte Carlo (MCMC). The presumed
phenotypic diversity. Patterns of correlations between number of clusters (optimal K) was then determined
pairwise of quantitative and qualitative traits was esti- using the software STRU​ CTU​ RE HARVESTER
mated separately via Pearson (r) and Cramer’s V (v) (Earl and VonHoldt 2012). Using the R package
coefficients, with P < 0.05 set as significance level, StructuRly 0.1.0 (Criscuolo and Angelini 2020), the
using packages "corrplot" (Wei et al. 2017) and "vcd" Bayesian clusters of the studied accessions were visu-
(Meyer et al. 2021), respectively. alised in the form of admixture bar plots. The genetic
A multifactorial analysis (MFA) was conducted relationship between accessions was visualised via
using the "FactoMineR" (Lê et al. 2008) and "fac- an unweighted pair group with arithmetic aver-
toextra" packages (Kassambara and Mundt 2020) to age (UPGMA) clustering using Poptree2 (Takezaki
examine patterns of the agro-morphological differ- et al. 2010) based on the matrix of Nei genetic dis-
entiation and determine the factors and quantitative tance between accessions pairwise generated by
and qualitative agro-morphological traits that explain GenAlEx 6.5.2 software (Peakall and Smouse 2006).
variability among accessions. Gower distance (Gower Genetic relatedness among kale accessions was fur-
1971) was calculated using the "daisy" function in the ther analysed using UPGMA (Unweighted Pair
"cluster" package (Maechler et al. 2012) to create a Group Method Arithmetic Average) cluster analysis
dissimilarity matrix between accessions based on based on Nei genetic distance, and clustering accu-
all traits, which was eventually used to visualise the racy was tested using 1000 bootstrap resamples with

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1225

the R package "poppr" (Kamvar et al. 2014). Finally, 17 accessions was 200–400 ­cm2 with a mean value of
the matrix of Nei’s genetic distance between acces- 261.94 ­cm2 and a CV of 42.42%. Phenotypic evalu-
sions pairwise was used along with the dissimilarity ation of leaf blade showed that leaf blade length
matrix between accessions based on Gower distance ranged from 16 to 21 cm in 13 accessions, with an
to perform a Mantel test with the package "vegan" average of 19.19 cm and a CV of 19.07% for all col-
(Oksanen et al. 2019) to test the degree of association lection. Leaf blade width was ≤ 15 cm in 8 acces-
between agro-morphological and molecular variation sions and 18–19 cm in 8 accessions with an aver-
of the studied kale accessions. age of 16.77 cm and a CV of 22.68%, while most
accessions (16) had a width/length ratio above the
average (0.46) with a CV of 17.26%. As for peti-
Results ole, length in most accessions was between 14 and
19 cm with an average of 17.02 cm (CV = 32.07%),
Morpho‑agronomic variability of kales from width in 15 accessions was above average (1.14) with
South‑East Europe CV of 25.19%, while length/width ratio in 14 acces-
sions was above average (16.06) with CV of 39.86%.
A total of 26 quantitative and qualitative morpho- Correlations between pairwise quantitative traits
agronomic descriptors related to plants, leaves, leaf (Fig. 3A) showed that, with the exception of the ratio
blades, and petioles were evaluated for these 26 kale between plant, leaf blade, and petiole, most of the
accessions originating from South-East Europe. remaining traits had strong positive associations with
Descriptive statistics revealed that the 11 quantita- each other (r = 0.39, P < 0.05). The highest correla-
tive traits had moderate (10% ≤ CV ≤ 20%) to very tions were observed between leaf and petiole length
high (CV ≥ 30%) morpho-agronomic variation for all (r = 0.86), leaf area and the blade width (r = 0.82),
traits with an average coefficient of variation (CV) plant diameter and leaf length (r = 0.79), and plant
of 25.33% (Table 1). The frequencies of distribution diameter and leaf petiole length (r = 0.76), most likely
of the 26 analysed accessions for the 11 quantitative reflecting the strong developmental associations
traits are shown in Fig. 1. Plant height was ≤ 60 cm among traits related to plant vigour. Interestingly,
in 16 accessions with a mean value of 61.34 cm and analysis of variance revealed a significant difference
a CV of 24.03%, plant diameter exceeded 60 cm in (P < 0.0001) among accessions for all quantitative
18 accessions with an average of 57.97 cm and a traits. As shown in Fig. 4A, variation between acces-
CV of 19.46%, while height/diameter was above the sions far exceeded variation within accessions, with
average (1.09) in 11 accessions with a CV value of a level of variation greater than 50% and an average
17.38%. As for leaf characteristics, the length in most of 73.98% for all traits. In addition, as indicated by
accessions (18) was 31–40 cm with a mean value the Shannon–Weaver index (H’), a wide variation was
of 36.39 cm and a CV of 19.22%, while the area in observed in all qualitative morphological descriptors.

Table 1  Summary of Plant organ Trait Min Max Mean Variance (n) SD (n) CV (%)
descriptive statistics of the
11 quantitative traits of the PLANT PLANTHEIGHT 37.3 111.5 61.34 217.26 14.74 24.03%
studied kale accessions
PLANTDIAM 26.3 88.5 57.97 127.21 11.28 19.46%
PLANTRATIO 0.86 1.61 1.09 0.04 0.19 17.38%
LEAF LEAFLENGTH 16.4 51.3 36.39 48.93 6.99 19.22%
LEAFAREA 74 640.4 261.94 12346.4 111.11 42.42%
LEAF BLADE BLADELENGHT 12.1 27.1 19.19 13.4 3.66 19.07%
BLADEWIDTH 7.1 23.8 16.77 14.46 3.8 22.68%
BLADERATIO 0.25 0.63 0.46 0.01 0.08 17.26%
Min minimum, Max PETIOLE PETIOLELENGTH 2.5 29.9 17.02 29.81 5.46 32.07%
maximum, SD Standard PETIOLEWIDTH 0.6 1.8 1.14 0.08 0.29 25.19%
deviation, CV coefficient of PETIOLERATIO 2.01 29.73 16.06 40.96 6.4 39.86%
variation

Vol.: (0123456789)
13
 1226 Genet Resour Crop Evol (2024) 71:1221–1239

Fig. 1  Frequency distribution of the 26 kale accessions for blade width [BLADEWIDTH]. h Leaf blade width/leaf length
the 11 quantitative traits. a Plant height [PLANTHEIGHT]. b ratio [BLADERATIO]. i Petiole length [PETIOLELENGTH].
Plant diameter [PLANTDIAM]. c Plant height/diameter ratio j Petiole width [PETIOLEWIDTH]. k Petiole length/width
[PLANTRATIO]. d Leaf length [LEAFLENGTH]. e Leaf area ratio [PETIOLERATIO]
[LEAFAREA]. f Leaf blade length [BLADELENGTH] g Leaf

For instance, plant shape (H’ = 0.69) was either was mostly obovate (11) or oblong (5), while blade
inverted pyramid or a dome (13 accessions). In the blistering (H’ = 1.54) was mostly low (11) or interme-
latter, the position of the growing point (H’ = 0.61) diate (7) (Fig. 2). In addition, anthocyanin (H’ = 0.68)
was deeply below the plant top for 10 accessions and was visible in the leaves of only 11 accessions, 10 of
slightly below in only 3 accessions (Fig. 2). Regard- which had partial distribution in the leaf (H’ = 0.49).
ing leaf shape, the leaf margin (H’ = 1.41) was mostly In most accessions, leaf colour (H’ = 1.03) was
crenate and serrate (9 and 6 accessions, respectively), mostly green (13) or dark green (9) and intensity
while the incision (H’ = 1.02) was mostly lyrate and (H’ = 1.01) was mostly medium (12) or dark (10).
sinuate (14 and 9 accessions, respectively). Moreo- Petiole colour (H’ = 1.21) was mostly light green (9),
ver, most accessions exhibited either a rounded (11) green (8), or purple (8) (Fig. 2), while petiole and/or
or broadly rounded (8) leaf apex (H’ = 1.07), the tip midvein enlargement (H’ = 0.82) was either interme-
was mostly dropping (17) (H’ = 0.76), and the lamina diate (13) or narrow (12) in most accessions. Moreo-
was mostly either concave drooping (12) or straight ver, as revealed by the Cramer’s coefficient (Fig. 3B),
(12) lamina (H’ = 0.91). The blade shape (H’ = 1.79) plant shape and growing point position, as well as

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1227

Fig. 2  Frequency distribution of the 26 kale accessions for h Leaf blade shape [BLADESHAPE]. i Leaf blade blistering
the 15 qualitative traits. a Plant shape [PLANTSHAPE]. b [BLADEBLISTER]. j Anthocyanin coloration of leaf [LEA-
Position of growing point in relation to top of plant (only for FANTHO]. k Distribution of anthocyanin coloration on leaf
those of dome shape) [PLANTPOSIT]. c Leaf division (mar- [LEAFDISTRIB]. l Leaf colour [LEAFCOLOUR]. m Intensity
gin) [LEAFMARGIN]. d Leaf division (incision) [LEAFIN- of colour of fully developed leaf [LEAFINTENS]. n Petiole
CISION]. (e) Leaf apex shape [LEAFAPEX]. f Leaf tip atti- and/or midvein enlargement [PETIOLEMEDVEIN]. o Petiole
tude [LEAFTIP]. g Leaf lamina attitude [LEAFLAMINA]. and/or midvein colour [PETIOLECOLOUR]

anthocyanin coloration and its distribution in the leaf, among kale accessions, data from both quantita-
were completely interdependent (r = 1). Some other tive and qualitative traits as well as the morphotype
traits appeared to be strongly (r ≥ 0.75) interdepend- and country of origin were analysed together using
ent, such as leaf colour and its intensity, as well as multifactorial analysis (MFA). The first three MFA
the blade shape and the distribution of anthocyanin components cumulatively explained 43.49% of the
coloration on the leaf. In contrast, no association total variance. As shown in Fig. 4B, the traits that
was detected between leaf anthocyanin coloration contributed most to the variation explained by the
and plant shape, and a weak association was found first three MFA dimensions, and therefore generally
between leaf apex shape and anthocyanin coloration differed among the studied accessions, were the leaf
and its distribution on the leaf (r ≤ 0.15). margin (22.08%), leaf apex (20.85%), intensity of col-
our of fully developed leaf (19.72%), leaf blade blis-
Patterns of the agro‑morphological differentiation tering (16.38%) and leaf incision (15.53%). Among
among accessions the quantitative traits, the petiole ratio (11.14%)
and the leaf area (10.99%) and the petiole width
To examine patterns of the agro-morphological vari- (10.04%) were the main contributors to differentia-
ation and determine the major sources of variation tion. Overall, the contribution of qualitative traits to

Vol.: (0123456789)
13
 1228 Genet Resour Crop Evol (2024) 71:1221–1239

Fig. 3  Patterns of correlations between pairwise quantitative A and qualitative traits B used for agro-morphological characterization
of the 26 kale accessions

the agro-morphological differentiation of the stud- agro-morphological diversification of kale accessions,

ied accessions slightly exceeded that of quantita- contributing with 32.61% (Fig. 4B). This pattern
tive traits (11.28% and 10.88%, respectively). How- was evidenced by the 2D plot of the first two MFA
ever, morphotype seems to be the main driver of dimensions, which showed that the two Slovenian

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1229

Fig. 4  A Variation decom-

position within and among
accessions for the 11 quan-
titative traits and B the
contribution of morphotype,
the 15 qualitative (filled bar
plots) and the 11 quantita-
tive (non-filled bar plots)
agro-morphological traits to
the total variance explained
by the first three dimen-
sions of the multifactorial
analysis

commercial cultivars of Scotch Kale (KIS19_R1) and (GD = 0.22) were the most phenotypically similar,
Marrow stem Kale (KIS19_R3) and the Italian com- with an average GD of 0.45 between all kale acces-
mercial cultivar of B. oleracea L. (KIS19_R2) formed sions. Cluster analysis based on Gower distance
three outgroups clearly separated from the other kale divided the 26 accessions into 8 groups (Fig. 5C).
accessions (Fig. 5A). When only the accessions of B. Consistent with the MFA results, accessions KIS19_
oleracea L. var. acephala DC. were examined, the 2D R1, KIS19_R2, and KIS19_R3 diverged from the
MFA plot did not show a clear separation between the other accessions in three different groups (C6, C1,
Croatian and Bosnian kales (Fig. 5B), suggesting that and C2). Cluster C2 included only Bosnian acces-
some variation may occur between geographically sions, while the four remaining clusters were a mix-
close sites within the same country. ture of Croatian and Bosnian accessions (Fig. 5D).
To further investigate the agro-morphological dif-
ferentiation among kale accessions, the degree of Microsatellite polymorphism and genetic diversity
dissimilarity was estimated using Gower distance level
(GD) based on phenotypic data of all traits. The
results showed that KIS19_R1 and KIS19_R3 were The twelve SSR markers used to characterise the
the most divergent accessions (GD = 0.87), while the genetic diversity of the 26 kale accessions yielded
two Bosnian accessions KIS19_R16 and KIS19_R26 a total of 108 alleles, with the observed number

Vol.: (0123456789)
13
 1230 Genet Resour Crop Evol (2024) 71:1221–1239

Fig. 5  Two-dimensional multifactorial analysis plots show- number of clusters and D the UPGMA dendrogram depicting
ing the relationship patterns between all 26 kale accessions the relationship between the 26 kale accessions based on the
(A) and only the B. oleracea var. acephala accession based on Gower distance matrix between pairwise accessions
the data of the 26 agro-morphological traits(B). C The optimal

of alleles (Na) ranging from three (Na14-C12) to of observed (Na) and effective (Ne) alleles was
14 (Ol10-A03a and Ol13-C12), with an average observed in accessions KIS19_R21 and KIS19_
of nine alleles per locus. The effective number of R27 (Na = 1.750; Ne = 1,750), whereas the highest
alleles (Ne) ranged from 2.01 to 3.16 in Na14-C12 allelic richness was observed in accessions KIS19_
and Ol10-A03a, respectively, with a mean of 2.62. R1 (Na = 3.667; Ne = 3.264), KIS19_R5 (Na = 3.5;
The 12 SSR markers had a high polymorphism rate, Ne = 3.160), and KIS19_R11 (Na = 3.667;
with a polymorphism information content (PIC) Ne = 3.124). Across all analysed accessions, the
of 0.4 (Na14-C12) to 0.86 (Ol12-F02) and a mean mean values for Na and Ne were 2.881 and 2.616,
PIC of 0.68. Major allele frequencies were lowest in respectively. Moreover, the Shannon diversity
Ol12-F02 (MAF = 0.27) and highest in Na12-C08 index (I) and expected heterozygosity (He), with
(MAF = 0.54), with a mean frequency of 0.39. In mean values of 0.953 and 0.578 across accessions,
addition, observed heterozygosity (Ho) varied from respectively, indicated that accessions KIS19_R11
0.76 (Ol13-C12) to 1 (Ra2-E11, Ol12-F11, and Ra2- were the most genetically diverse (I = 1.887 and
E03) with a mean of 0.95. Importantly, high gene He = 0.662), whereas accessions KIS19_R21 and
diversity expressed by expected heterozygosity (He) KIS19_R27 were the least diverse (I = 0.587 and
was observed in all SSR markers, with values ranging He = 0.417). Remarkably, the observed heterozy-
from 0.52 in Na14-C12 to 0.87 in Ol12-F02 and He gosity was higher than the expected heterozygosity
of 0.73 in all markers. This pattern is reflected in the in all accessions, with an average value of 0.939,
negative Fis values obtained for all markers, indicat- resulting in negative inbreeding coefficients (Fis)
ing a general excess of heterozygosity (Table 2). that averaged -0.671, confirming the high level of
Genetic diversity estimates obtained at the acces- heterozygosity in the entire collection.
sion level are shown in Table 3. The lowest number

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1231

Table 2  Genetic diversity Loci Na Ne MAF PIC Ho He Fis

parameters of the 12 simple
sequence repeat (SSR) Na12-C08 10 2.091 0.535 0.395 0.944 0.517 − 0.895
markers used for molecular
Na14-C12 3 3.005 0.500 0.521 0.986 0.604 − 0.848
characterization of kale
accessions Ol10-A03a 14 2.349 0.317 0.545 0.958 0.621 − 0.809
Ol10-F11a 6 2.005 0.479 0.648 0.930 0.677 − 0.660
Ol10-H02 5 2.224 0.493 0.650 1.000 0.708 − 0.751
Ol11-G11 7 3.161 0.331 0.710 0.972 0.755 − 0.533
Na number of alleles Ol11-H02 11 2.971 0.366 0.720 1.000 0.767 − 0.573
observed, Ho observed Ol12-F02 16 2.187 0.275 0.749 1.000 0.780 − 0.593
heterozygosity, MAF major Ol12-F11 5 3.026 0.289 0.775 0.986 0.802 − 0.537
allele frequency, PIC
Ol13-C12 14 3.025 0.366 0.831 0.972 0.851 − 0.483
polymorphism Information
Content, Ho observed Ra2-E03 12 2.883 0.387 0.836 0.756 0.860 − 0.437
heterozygosity, He expected Ra2-E11 5 2.460 0.394 0.856 0.901 0.873 − 0.460
heterozygosity, Fis fixation Mean 9 2.616 0.394 0.686 0.950 0.735 − 0.632
index

Table 3  Sample size Accession N Na Ne Ho I He Fis No. PA

(N) and genetic diversity
estimates among the 26 kale KIS19_R1 3 3.667 3.264 0.944 1.171 0.648 − 0.491 0.083
accessions
KIS19_R2 3 3.167 2.751 0.931 1.032 0.605 − 0.566 0.083
KIS19_R3 2 2.750 2.722 1 0.953 0.594 − 0.744 0.083
KIS19_R4 3 3.083 2.845 0.917 1.026 0.611 − 0.554 0.083
KIS19_R5 3 3.500 3.160 0.972 1.129 0.639 − 0.565 0
KIS19_R6 3 3.250 2.885 0.917 1.072 0.625 − 0.498 0
KIS19_R8 2 2.000 2.000 1 0.693 0.5 −1 0
KIS19_R9 3 3.083 2.842 1 1.009 0.602 − 0.728 0
KIS19_R10 2 2.250 2.133 0.875 0.769 0.545 − 0.77 0.083
KIS19_R11 3 3.667 3.124 1 1.187 0.662 − 0.532 0
KIS19_R12 3 3.333 3.023 0.972 1.088 0.629 − 0.605 0.083
KIS19_R13 2 2.083 2.000 0.917 0.722 0.49 − 0.891 0
KIS19_R14 3 2.833 2.653 1 0.946 0.584 − 0.766 0.167
KIS19_R15 3 2.917 2.575 0.944 0.957 0.579 − 0.647 0
KIS19_R16 3 2.833 2.546 0.958 0.94 0.578 − 0.682 0.083
KIS19_R17 3 3.000 2.636 0.972 1.007 0.606 − 0.618 0
KIS19_R18 3 3.250 2.904 0.972 1.072 0.625 − 0.591 0
KIS19_R19 3 3.083 2.637 0.917 0.995 0.588 − 0.583 0.083
KIS19_R20 3 2.583 2.354 0.917 0.86 0.532 − 0.768 0.167
KIS19_R21 2 1.750 1.750 0.833 0.578 0.417 −1 0
Na number of alleles KIS19_R22 3 3.333 2.888 0.958 1.098 0.635 − 0.533 0
observed, Ne effective KIS19_R23 3 3.250 2.776 0.833 1.040 0.593 − 0.418 0.167
number of alleles, Ho KIS19_R24 3 2.750 2.510 0.958 0.921 0.573 − 0.7 0.083
observed heterozygosity,
KIS19_R25 2 2.667 2.522 0.917 0.913 0.573 − 0.633 0
I Shannon’s information
index, He expected KIS19_R26 3 3.083 2.755 0.944 1.033 0.616 − 0.563 0
heterozygosity, Fis fixation KIS19_R27 2 1.750 1.750 0.833 0.578 0.417 −1 0
index, No. PA number of Mean 2.731 2.881 2.616 0.939 0.953 0.578 − 0.671 0.048
private alleles

Vol.: (0123456789)
13
 1232 Genet Resour Crop Evol (2024) 71:1221–1239

Genetic differentiation and population structure any geographic grouping patterns. As shown in
Fig. 7, the 26 kale accessions could be divided into
Analysis of molecular variance (AMOVA) revealed two main groups with 100% bootstrap support. The
that 94% of the genetic variation occurred within first group included six Bosnian accessions, while
accessions. Genetic differentiation among acces- the 20 remaining accessions were mixed in a second
sions (Fst) ranged from 0.01 to 0.294, with an aver- major group that was divided into several subgroups
age of 0.06 and a level of gene flow of 3.55 migrants differentiated with low bootstraps support. Within
per generation. However, no genetic differentiation the second group, the three commercial cultivars
was observed between countries of origin (Fst = 0). were separated from the remaining accessions in a
Bayesian clustering using STRU​ CTU​RE revealed single group.
that ΔK reached its maximum (ΔK = 3.84) at K = 4,
indicating that the accessions could be divided into
four main clusters (Fig. 6A) without a geographic Association between agro‑morphological and
clustering pattern. With the exception of the Ital- molecular variation
ian commercial cultivar (KIS19_R2), the Croatian
accession KIS19_R9, and the three Bosnian acces- The Mantel test revealed a very low and non-signif-
sions KIS19_R20, KIS19_R25, and KIS19_R27, icant association between the Nei’s genetic distance
which were assigned to cluster 3 with a member- matrix generated from the molecular data and the
ship coefficient Q > 75%, the remaining accessions Gower distance matrix among accessions, with a
exhibited mixed clustering with partial member- cophenetic coefficient (r) of 0.06 (P = 0.213). When
ship in multiple clusters (Fig. 6B). Remarkably, only the qualitative trait data were used, r was 0.06
very little divergence in allele frequency (DAF) was (P = 0.249) and dropped to 0.03 (P = 0.520) when
detected among the four clusters (< 5%), with com- only the quantitative traits were used. This result
pletely identical genetic composition in Cluster 2 suggests that molecular differentiation is not cor-
and Cluster 4 (DAF = 0), indicating little genetic related with agro-morphological differentiation
differentiation among kale accessions. Similar to among kale accessions, as clearly evidenced by the
the STRU​CTU​RE based clustering, UPGMA clus- different patterns of clustering in our study.
tering based on Nei’s genetic distance did not reveal

Fig. 6  A A graph showing

the relationship between
ΔK and K values; B the
inferred genetic structure
of the 26 kale acces-
sions based on Bayesian
clustering of the 12 simple
sequence repeat (SSR)
marker data

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1233

Fig. 7  UPGMA clustering

showing the relationship
between the 26 kale acces-
sions based on Nei’s genetic
distances derived from the
12 simple sequence repeat
(SSR) markers data. Values
in the branches indicate the
clustering support at 1000
bootstraps resamples

Discussion SSR markers was combined to characterise the

genetic variation of 23 Croatian and Bosnian leafy
Detailed genetic characterization of a particular kale accessions for inclusion in future breeding pro-
germplasm is instrumental for establishing appro- grammes. Similar to previous research on Spanish
priate management, preservation and breeding (Cartea et al. 2003) and Indian (Gorka et al. 2018;
practices. Within the Brassicaceae family, B. olera- Singh et al. 2017) accessions, our study showed con-
cea var. acephala L. (leafy kale) has attracted great siderable agro-morphological variation in quantita-
interest worldwide as a ’superfood’ and ornamental tive traits among the studied kale accessions. Despite
plant (Šamec et al. 2018). Therefore, several investi- the similarity in leaf morphology in terms of length
gations have already been conducted to characterize and width, the Indian kale accessions had shorter
the genetic diversity of local accessions in different plant height (average 47.71 cm versus 61.34 cm)
countries based on agro-morphological or nutritional compared to the accessions analysed in our study
traits, e.g. in Portugal (Dias et al. 1994), Spain (Car- (Singh et al. 2017). In contrast, the Spanish acces-
tea et al. 2003; Padilla et al. 2007), Turkey (Balkaya sions were characterised by outstanding plant vigour
and Yanmaz 2005), Croatia (Šamec et al. 2019a) and with plant height exceeding 160 cm. Interestingly,
Italy (Lotti et al. 2018). Some other investigations most of the quantitative traits analysed in our study
have been conducted using molecular markers such as showed strong positive correlation among themselves,
random amplified polymorphic DNA (RAPD) (Farn- indicating polygenic control through the pleiotropic
ham 1996; Okumus and Balkaya 2007), amplified effect of genes on different traits or linkage disequi-
fragment length polymorphism (AFLP) (Christensen librium among different loci (Chebib and Guillaume
et al. 2011), single nucleotide polymorphism (SNP) 2021). Similar results were also observed in B. oler-
(Hahn et al. 2022) and SSR markers (Šutković et al. acea L. var. botrytis (Kumar et al. 2019), B. rapa
2021). In addition to these valuable reports, detailed var. rapa L. (Khadivi et al. 2022) and B. carinata
characterization of kale genetic resources will be very A. Braun (Yimer et al. 2021) accessions. Moreover,
useful to breed superior accessions, promote their our results revealed that most of the phenotypic vari-
valorisation and management to halt genetic erosion ation in quantitative traits resided among accessions.
and economic losses due to climate change. Overall, the strong correlations among traits and the
In the current study, a large number of qualita- large phenotypic diversity among accessions provide
tive and quantitative agro-morphological traits and a great opportunity to use both traits and accessions

Vol.: (0123456789)
13
 1234 Genet Resour Crop Evol (2024) 71:1221–1239

analysed in our study as a source of genetic diversity agro-morphological traits. Interestingly, in agreement
in future breeding programmes. In addition, a large with our results, previous research has shown that
phenotypic diversity was found in qualitative traits qualitative traits such as leaf colour and division, leaf
(H’ = 0.49 – 1.54). Similar to the Indian accessions blade blistering, anthocyanin coloration, and antho-
analysed by (Gorka et al. 2018), most of our acces- cyanin coloration distribution were the main drivers
sions were either an inverted pyramid or a dome of variation in Croatian accessions of B. oleracea var.
shaped with green leaves. In contrast to the low vari- acephala with different seed source (Batelja et al.
ation in leaf blade shape among Indian accessions, 2009; Šamec et al. 2019a).
which were mainly characterised by elliptic blade In addition to agro-morphological characteriza-
shape, this trait was most variable among the acces- tion, molecular characterization, particularly via
sions analysed in our study (H’ = 1.54). codominant markers such as microsatellites (SSRs),
Previous research has demonstrated that leaf is required to provide a robust appraisal of the genetic
shape is strongly influenced by environmental fac- diversity of kale accessions. The joint analysis maxi-
tors in addition to its genetic control (Dkhar and mizes the comprehensive assessment of the diversity
Pareek 2014) and that its variation may be the of available genetic material, facilitates its organiza-
result of strong adaptation to environmental stresses tion, and enables the identification of desired traits
(Kidner and Umbreen 2010). Šamec et al. (2019a) and distinctive alleles, allowing the accurate selection
reported that kale exhibits broad tolerance to differ- of promising genetic resources for breeding programs
ent temperatures. On the other hand, variation in leaf (Li et al. 2020). In our study, the twelve SSR markers
shape in kale is thought to be the result of selection used for genetic characterization of kale accessions,
by the breeder (Arias et al. 2021). Thus, the varia- previously shown to be highly informative (El-Esawi
tion in leaf shape observed in our study could be due et al. 2016), were found to be highly polymorphic
to the adaptation of the studied accessions to their with a mean PIC value (0.68) and an average of nine
local environment, since they originated from dif- alleles. These values were higher than those reported
ferent locations. Cartea et al. (2003) also considered for the same SSR set in Irish accessions (El-Esawi
that the agro-morphological variation among Span- et al. 2016) and for other SSR markers in different
ish kale accessions could be due to both adaptation to commercial cultivars of B. oleracea L. (Izzah et al.
local ecological conditions and selection by farmers. 2013; Tonguç and Griffiths 2004). This result indi-
In addition, the MFA analysis revealed that qualita- cates the high reliability of these SSR markers for the
tive rather than quantitative traits contributed most to analysis of the genetic structure of kale accessions.
agro-morphological diversification among the stud- Our results revealed that the accessions analysed
ied kale accessions. The MFA analysis and UPGMA in this study had high genetic diversity (He = 0.578),
clustering based on the Gower distance also revealed which exceeded the level of Irish kale (He = 0.459),
that the three commercial cultivars were classified as cabbage (He = 0.548), cauliflower (He = 0.446) and
independent outliers that deviated from the other kale sprouts (He = 0.536) using SSR markers (El-Esawi
accessions, with geographic origin being an impor- et al. 2016). Moreover, similar to the above study, a
tant contributor to diversification. A similar pattern significant excess of heterozygosity was observed in
of agro-morphological differentiation was observed our study as reflected by the high observed heterozy-
in cabbage and Galega kale landraces from Portu- gosity and negative inbreeding coefficient (Fis). This
gal (Dias et al. 1994) and Spanish kale populations was to be expected considering the small number of
(Cartea et al. 2003). The considerable effect of geo- individuals within the studied accessions (2 to 3 indi-
graphic origin on differentiation and the evidence of viduals) and the outcrossing breeding system of B.
adaptation to the local environment suggest that the oleracea L.
influence of genotype by environment on several On the other hand, in contrast to the results
agro-morphological traits is important. Moreover, we obtained for quantitative traits, the SSR data
the divergence among the three commercial culti- revealed that most of the genetic variation occurred
vars with each other and with the kale accessions at the level within accessions (95%), indicating low
was expected, as they belong to different morpho- genetic differentiation and weak genetic structure
types and indicate that they were selected for different in the studied accessions (Fst = 0.05). Although

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1235

genetic differentiation was greater, a similar pattern and non-overlap between the agro-morphological and
was observed in Irish kale (Fst = 11%), cauliflower genotypic data to investigate the extent and nature
(Fst = 11%), and Brussels sprouts (Fst = 9%) (El- of genetic variability in crops (Da Silva et al. 2017;
Esawi et al. 2016), as well as in closely related cul- Darkwa et al. 2020; Geleta et al. 2006). Importantly,
tivars such as B. oleracea L. var. italica (Fst = 13%) this pattern indicates that the polymorphism detected
(Shen et al. 2021) and B. rapa subsp. rapa L by SSR markers does not fully contribute to the agro-
(Fst = 10%) (Soengas et al. 2011). The low genetic morphological variability among kale accessions,
differentiation observed in our study was supported suggesting that only a few genes are involved in the
by the relatively high gene flow observed between desired agro-morphological traits. Another plausible
accessions (Nm = 3.55), which was higher than the explanation could be the massive influence of geno-
level observed for B. rapa subsp. rapa (Nm = 2.25) type-environment interaction on the quantitatively
(Soengas et al. 2011) and B. oleracea L. var. ital- inherited complex agro-morphological traits con-
ica (Nm = 1.74) (Shen et al. 2021). Such excellent trolled by several genomic regions, including SSR
gene flow can occur naturally or via seed exchange loci which were not used in the present study.
between farmers given the open pollination in B. oler-
acea L. In agreement with the above research, as well
as with the study of Shen et al. (2021) on B. olera- Conclusion
cea L. var. italica, the weak genetic structure of the
studied accessions was confirmed by UPGMA clus- In conclusion, the present study presented a detailed
tering and Bayesian analysis, which did not reveal assessment of the genetic diversity and structure of
a clear relationship between clustering of acces- kale accessions using a large set of agro-morpholog-
sions and geographical origin. A similar pattern was ical traits and SSR markers. The results revealed that
also observed in wild populations of B. oleracea L. the kale accessions exhibit wide variation in several
(Christensen et al. 2011; Lanner-Herrera et al. 1996; economically important agro-morphological traits.
Mittell et al. 2020) and B. oleracea L. var. italica The agro-morphological characterization revealed
(Shen et al. 2021). In contrast, in a large collection that qualitative traits contributed more than quan-
of Spanish B. oleracea L. wild accessions, restricted titative traits to phenotypic differentiation among
gene flow (Nm = 0.26) and large genetic differentia- accessions. The diversity among accessions in some
tion (Fst = 55.1%) were observed, mainly promoted particular traits such as leaf shape, leaf colour and
by geographic isolation (Tortosa et al. 2017). How- distribution of anthocyanin coloration on the leaf is
ever, in our study, UPGMA clustering showed that useful to select a subset of accessions for breeding
the three commercial standard cultivars, although programs. The SSR markers used in our study, which
belonging to three different subspecies, were clearly proved to be very useful for molecular charcteriza-
differentiated from kale accessions in a single group, tion, showed great genetic diversity at the molecu-
suggesting that their genomes harbour common lar level. As expected, the SSR data showed a large
genomic regions and could be bred for similar agro- intra-accession variation and a weak genetic structure
nomic traits. Moreover, our results revealed that the non-associated with the geographic origin of the kale
grouping of accessions based on the agro-morpho- accessions. Importantly, the weak and nonsignificant
logical data did not match the grouping determined association between dissmilarity matrices among
from the SSR data. This was confirmed by the weak accessions generated by agro-morphological and SSR
and non-significant association between the dissimi- data indicates the complementarity of the multidis-
larity matrices between the accessions determined ciplianry approaches used in our study. Overall, the
by the Mantel test for both pairs of data. The weak results of our study pointed to the importance of the
association between the phenotypic and genotypic kale collection studied as a valuable reservoir of agro-
matrices was also reported in broccoli (Li et al. 2019) morphological and genetic variability that could be
and cabbage (Kang et al. 2011) and several other exploited in future breeding programs.
crops (Hartings et al. 2008; Roldan-Ruiz et al. 2000;
Soriano et al. 2016). Previous research indicated that Acknowledgements This study was part of the research pro-
such incongruence highlights the complementarity gramme “Agrobiodiversity” (P4-0072) financially supported

Vol.: (0123456789)
13
 1236 Genet Resour Crop Evol (2024) 71:1221–1239

by the Slovenian Research Agency, Ljubljana, Slovenia; part References

of the EUBRASWILD project funded by the international
ECPGR project (Third Call – Phase X 2019-2023); and part Arias T, Niederhuth CE, McSteen P, Pires JC (2021) The
of the bilateral projects funded by ARRS: Slovenia-Serbia molecular basis of kale domestication: transcriptional pro-
(BI-RS/20-21-015) and Slovenia-Bosnia and Herzegovina filing of developing leaves provides new insights into the
(BI-BA/19-20-009 and BI-BA/21-23-011). The authors thank evolution of a Brassica oleracea vegetative morphotype.
Boštjan Ogorevc, Mojca Polak, Neža Kermavnar, Ana Špilak, Front Plant Sci 12:637115. https://​doi.​org/​10.​3389/​fpls.​
and Teja Krpan for technical assistance, evaluations, and analy- 2021.​637115
ses. The authors thank the Genetic Resources Institute, Univer- Balkaya A, Yanmaz R (2005) Promising kale (Brassica olera-
sity of Banja Luka (Bosnia and Herzegovina) and the Institute cea var. acephala) populations from Black Sea region,
for Adriatic Crops and Karst Reclamation, Split (Croatia) for Turkey. New Zeal J Crop Hort Sci 33:1–7. https://​doi.​org/​
providing accessions from their gene banks. 10.​1080/​01140​671.​2005.​95143​24
Batelja K, Goreta Ban S, Žanić K, Miloš B, Dumičić G, Mato-
Author contributions All authors contributed to the study tan Z (2009) Autochtonous kale populations (Brassica
conception and design. Material preparation, data collec- oleracea L. var. acephala) in Croatian coastal region. Pol-
tion and analysis were performed by BP, LS and MN. The joprivreda 15:8–14
first draft of the manuscript was written by LS and all authors Cartea ME, Picoaga A, Soengas P, Ordás A (2003) Morpho-
commented on previous versions of the manuscript. VM logical characterization of kale populations from north-
obtained the funding. All authors read and approved the final western Spain. Euphytica 129:25–32. https://​doi.​org/​10.​
manuscript. 1023/A:​10215​76005​211
Chebib J, Guillaume F (2021) Pleiotropy or linkage? Their rel-
Funding This study was part of the research programme ative contributions to the genetic correlation of quantita-
“Agrobiodiversity” (P4-0072) financially supported by the tive traits and detection by multitrait GWA studies. Genet-
Slovenian Research Agency, Ljubljana, Slovenia; part of the ics 219:iyab159. https://​doi.​org/​10.​1093/​genet​ics/​iyab1​59
EUBRASWILD project funded by the international ECPGR Choudhury A, Deb S, Kharbyngar B, Rajpal VR, Rao SR
project (Third Call – Phase X 2019-2023); and part of the bilat- (2022) Dissecting the plant genome: through new genera-
eral projects funded by ARRS: Slovenia-Serbia (BI-RS/20–21- tion molecular markers. Genet Resour Crop Evol 69:2661–
015) and Slovenia-Bosnia and Herzegovina (BI-BA/19-20-009 2698. https://​doi.​org/​10.​1007/​s10722-​022-​01441-3
and BI-BA/21-23-011). Christensen S, von Bothmer R, Poulsen G, Maggioni L, Phillip
M, Andersen BA, Jørgensen RB (2011) AFLP analysis of
Data availability The authors confirm that all the experi- genetic diversity in leafy kale (Brassica oleracea L. con-
mental data are available and accessible via the main text and/ var. acephala (DC.) Alef.) landraces, cultivars and wild
or the supplementary information. Raw data will be made populations in Europe. Genet Resour Crop Evol 58:657–
available on request. 666. https://​doi.​org/​10.​1007/​s10722-​010-​9607-z
Comtois D, Comtois MD (2016) Package ‘summarytools.’
Declarations CPVO (2011) Protocol for Distinctness, Uniformity and Stabil-
ity tests: Brassica oleracea L. var. sabellica L. Curly kale:
Conflict of interest The authors declare that there are no con- 1–15. Available online: https://​cpvo.​europa.​eu/​sites/​defau​
flicts of interest. All the authors read and approved the manu- lt/​files/​docum​ents/​phase​olus_​vulga​ris .pdf
script in its final form. Criscuolo NG, Angelini C (2020) StructuRly: A novel shiny
app to produce comprehensive, detailed and interac-
Open Access This article is licensed under a Creative Com- tive plots for population genetic analysis. PLoS ONE
mons Attribution 4.0 International License, which permits 15:e0229330. https://​doi.​org/​10.​1371/​journ​al.​pone.​02293​
use, sharing, adaptation, distribution and reproduction in any 30
medium or format, as long as you give appropriate credit to the Da Silva AR, Malafaia G, Menezes IPP (2017) Biotools: an R
original author(s) and the source, provide a link to the Crea- function to predict spatial gene diversity via an individual-
tive Commons licence, and indicate if changes were made. The based approach. Genet Mol Res 16:1–6. https://​doi.​org/​10.​
images or other third party material in this article are included 4238/​gmr16​029655
in the article’s Creative Commons licence, unless indicated Darkwa K, Agre P, Olasanmi B, Iseki K, Matsumoto R, Pow-
otherwise in a credit line to the material. If material is not ell A, Bauchet G, De Koeyer D, Muranaka S, Adebola
included in the article’s Creative Commons licence and your P, Asiedu R, Terauchi R, Asfaw A (2020) Comparative
intended use is not permitted by statutory regulation or exceeds assessment of genetic diversity matrices and cluster-
the permitted use, you will need to obtain permission directly ing methods in white Guinea yam (Dioscorea rotundata)
from the copyright holder. To view a copy of this licence, visit based on morphological and molecular markers. Sci Rep
http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. 10:13191. https://​doi.​org/​10.​1038/​s41598-​020-​69925-9
Dias JS, Monteiro AA, Kresovich S (1994) Genetic diversity
and taxonomy of Portuguese Tronchuda cabbage and
Galega kale landraces using isozyme analysis. Euphytica
75:221–230. https://​doi.​org/​10.​1007/​BF000​25607

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1237

Dkhar J, Pareek A (2014) What determines a leaf’s shape? of genetic diversity in 91 commercial Brassica olera-
EvoDevo 5:1–19. https://​doi.​org/​10.​1186/​2041-​9139-5-​47 cea L. cultivars belonging to six varietal groups. Genet
Earl DA, von Holdt BM (2012) STRU​CTU​RE HARVESTER: Resour Crop Evol 60:1967–1986. https://​doi.​org/​10.​1007/​
a website and program for visualizing STRU​ CTU​ RE s10722-​013-​9966-3
output and implementing the Evanno method. Con- Kamvar ZN, Tabima JF, Grünwald NJ (2014) Poppr: an R
serv Genet Resour 4:359–361. https://​doi.​org/​10.​1007/​ package for genetic analysis of populations with clonal,
s12686-​011-​9548-7 partially clonal, and/or sexual reproduction. PeerJ 2:e281
El-Esawi MA, Germaine K, Bourke P, Malone R (2016) Kang J, Fang Z, Wang X, Xu D, Liu Y, Yang L, Zhuang M,
Genetic diversity and population structure of Brassica Zhang Y (2011) Genetic diversity and relationships
oleracea germplasm in Ireland using SSR markers. CR among cabbage (Brassica oleracea var. capitata) lan-
Biol 339:133–140. https://​doi.​org/​10.​1016/j.​crvi.​2016.​02.​ draces in China revealed by AFLP markers. Afr J Biotech-
002 nol 10:5940–5949
Farnham MW (1996) Genetic variation among and within Kassambara A, Mundt F (2020a) Extrac and visualize the
United States collard cultivars and landraces as deter- results of multivariate data analyses. R Package Version
mined by randomly amplified polymorphic DNA mark- 1.0. 3. R package version.
ers. J Am Soc Hortic Sci 121:374–379. https://​doi.​org/​10.​ Khadivi A, Mirheidari F, Moradi Y (2022) Selection of supe-
21273/​JASHS.​121.3.​374 rior accessions of turnip (Brassica rapa var. rapa L.)
Geleta N, Labuschagne MT, Viljoen CD (2006) Genetic based on tuber quality-related characters. Food Sci Nutr
diversity analysis in sorghum germplasm as estimated 10:2667–2680. https://​doi.​org/​10.​1002/​fsn3.​2871
by AFLP, SSR and morpho-agronomical markers. Bio- Kidner CA, Umbreen S (2010) Why is leaf shape so variable.
divers Conserv 15:3251–3265. https://​doi.​org/​10.​1007/​ Int J Plant Dev Biol 4:64–75
s10531-​005-​0313-7 Kuerban A, Yaghmoor SS, Almulaiky YQ, Mohamed YA,
Ghaffari P, Talebi R, Keshavarzi F (2014) Genetic diversity Razvi SSI, Hasan MN, Moselhy SS, Al-Ghafari AB,
and geographical differentiation of Iranian landrace, culti- Alsufiani HM, Kumosani TA (2017) Therapeutic effects
vars, and exotic chickpea lines as revealed by morphologi- of phytochemicals of brassicaceae for management of
cal and microsatellite markers. Physiol Mol Biol Plants obesity. J Pharm Res Int 19:1–11. https://​doi.​org/​10.​9734/​
20:225–233. https://​doi.​org/​10.​1007/​s12298-​014-​0223-9 JPRI/​2017/​37617
Golicz AA, Bayer PE, Barker GC, Edger PP, Kim H, Martinez Kumar V, Thakur RK, Kumar P (2019) Assessment of heavy
PA, Chan CKK, Severn-Ellis A, McCombie WR, Parkin metals uptake by cauliflower (Brassica oleracea var botry-
IA (2016) The pangenome of an agronomically impor- tis) grown in integrated industrial effluent irrigated soils:
tant crop plant Brassica oleracea. Nat Commun 7:13390. A prediction modeling study. Sci Hortic 257:108682.
https://​doi.​org/​10.​1038/​ncomm​s13390 https://​doi.​org/​10.​1016/j.​scien​ta.​2019.​108682
Gonçalves ÁLM, Lemos M, Niero R, de Andrade SF, Maistro Lanner-Herrera C, Gustafeson M, Filt AS, Bryngelsson T
EL (2012) Evaluation of the genotoxic and antigenotoxic (1996) Diversity in natural populations of wild Brassica
potential of Brassica oleracea L. var. acephala D.C. in oleracea as estimated by isozyme and RAPD analysis.
different cells of mice. J Ethnopharmacol 143:740–745. Genet Resour Crop Evol 43:13–23. https://​doi.​org/​10.​
https://​doi.​org/​10.​1016/j.​jep.​2012.​07.​044 1007/​BF001​26936
Gorka S, Samnotra RK, Kumar S, Chopra S, Gupta M (2018) Li Z, Mei Y, Liu Y, Fang Z, Yang L, Zhuang M, Zhang Y,
Analysis of genetic diversity in kale (Brassica oleracea L. Lv H (2019) The evolution of genetic diversity of broc-
var. acephala) genotypes of jammu and kashmir region coli cultivars in China since 1980. Sci Hortic 250:69–80.
based on morphological descriptors. Int J Curr Microbiol https://​doi.​org/​10.​1016/j.​scien​ta.​2019.​02.​034
Appl Sci 7:2176–2181. https://​doi.​org/​10.​20546/​ijcmas.​ Li D, Long C, Pang X, Ning D, Wu T, Dong M, Han X, Guo
2018.​702.​261 H (2020) The newly developed genomic-SSR markers
Gower JC (1971) A general coefficient of similarity and some uncover the genetic characteristics and relationships of
of its properties. Biometrics 27:857–871. https://​doi.​org/​ olive accessions. Plant Biol 8:e8573. https://​doi.​org/​10.​
10.​2307/​25288​23 7717/​peerj.​8573
Hahn C, Howard NP, Albach DC (2022) Different Shades of Liu K, Muse SV (2005) PowerMarker: an integrated analysis
Kale - Approaches to analyze kale variety interrelations. environment for genetic marker analysis. Bioinformatics
Genes 13:232. https://​doi.​org/​10.​3390/​genes​13020​232 21:2128–2129. https://​doi.​org/​10.​1093/​bioin​forma​tics/​
Hartings H, Berardo N, Mazzinelli GF, Valoti P, Verderio A, bti282
Motto M (2008) Assessment of genetic diversity and Lotti C, Iovieno P, Centomani I, Marcotrigiano AR, Fanelli
relationships among maize (Zea mays L.) Italian lan- V, Mimiola G, Summo C, Pavan S, Ricciardi L (2018)
draces by morphological traits and AFLP profiling. Genetic, bio-agronomic, and nutritional characterization
Theor Appl Genet 117:831–842. https://​doi.​org/​10.​1007/​ of kale (Brassica oleracea L. var. acephala) diversity in
s00122-​008-​0823-2 Apulia Southern Italy. Diversity 10:25. https://​doi.​org/​10.​
IBPGR (1990). Descriptors for Brassica and Raphanus. Inter- 3390/​d1002​0025
national Board for Plant Genetic Resources, Rome, Italy: Maechler M, Rousseeuw P, Struyf A, Hubert M, Hornik K
1–58. https://​cropg​eneba​nk.​sgrp.​cgiar.​org/​images/​file/​ (2012) Cluster: cluster analysis basics and extensions.
learn​ing_​space/​descr​iptors_​brass​ica_​rapha​nus .pdf Meglič V, Pipan B (2018) Spatial and temporal assessment of
Izzah NK, Lee J, Perumal S, Park JY, Ahn K, Fu D, Kim GB, Brassica napus L. maintaining genetic diversity and gene
Nam YW, Yang TJ (2013) Microsatellite-based analysis flow potential: An empirical evaluation. In: El-Esawi MA

Vol.: (0123456789)
13
 1238 Genet Resour Crop Evol (2024) 71:1221–1239

(ed) Brassica Germplasm - Characterization, Breeding Pritchard JK, Stephens M, Donnelly P (2000) Inference of pop-
and Utilization. InTech ulation structure using multilocus genotype data. Genetics
Mercati F, Longo C, Poma D, Araniti F, Lupini A, Mammano 155:945–959. https://​doi.​org/​10.​1093/​genet​ics/​155.2.​945
MM, Fiore MC, Abenavoli MR, Sunseri F (2015) Genetic R Core Team (2021). R: A language and environment for sta-
variation of an Italian long shelf-life tomato (Solanum tistical computing. R Foundation for Statistical Comput-
lycopersicon L.) collection by using SSR and morpho- ing, Vienna, Austria. URL https://​www.R-​proje​ct.​org/.
logical fruit traits. Genet Resour Crop Evol 62:721–732. Rakshita KN, Singh S, Verma VK, Sharma BB, Saini N, Ique-
https://​doi.​org/​10.​1007/​s10722-​014-​0191-5 bal MA, Sharma A, Dey SS, Behera TK (2021) Agro-
Meyer D, Zeileis A, Hornik K, Gerber F, Friendly M, Meyer morphological and molecular diversity in different matu-
MD (2021) Package ‘vcd.’ Visualizing Categorical Data. rity groups of Indian cauliflower (Brassica oleracea var.
R Package Version 1–4. botrytis L.). PLoS ONE 16:e0260246. https://​doi.​org/​10.​
Mittell EA, Cobbold CA, Ijaz UZ, Kilbride EA, Moore KA, 1371/​journ​al.​pone.​02602​46
Mable BK (2020) Feral populations of Brassica olera- Riangwong K, Wanchana S, Aesomnuk W, Saensuk C, Nuban-
cea along Atlantic coasts in western Europe. Ecol Evol koh P, Ruanjaichon V, Kraithong T, Toojinda T, Vanavi-
10:11810–11825. https://​doi.​org/​10.​1002/​ece3.​6821 chit A, Arikit S (2020) Mining and validation of novel
Oksanen J, Blanchet FG, Friendly M, Kindt R, Leg- genotyping-by-sequencing (GBS)-based simple sequence
endre P, McGlinn D, Imports MASS (2019). Package repeats (SSRs) and their application for the estimation of
‘vegan’. Community ecology package, version, 2(9) the genetic diversity and population structure of coconuts
Okumus A, Balkaya A (2007) Estimation of genetic diversity (Cocos nucifera L.) in Thailand. Hortic Res 7:156. https://​
among Turkish kale populations (Brassica oleracea var. doi.​org/​10.​1038/​s41438-​020-​00374-1
acephala L.) using RAPD markers. Russ J Genet 43:411– Rivera A, Monteagudo AB, Igartua E, Taboada A, García-
415. https://​doi.​org/​10.​1134/​S1022​79540​70400​96 Ulloa A, Pomar F, Riveiro-Leira M, Silvar C (2016)
Oumata S, Monneveux P, Zaharieva M, Mekliche-Hanifi L, Assessing genetic and phenotypic diversity in pepper
David J (2023) Variation of morphological traits among (Capsicum annuum L.) landraces from North-West Spain.
wheat (Triticum aestivum L.) landraces from two regions Sci Hortic 203:1–11. https://​doi.​org/​10.​1016/j.​scien​ta.​
of the Algerian Sahara. Potential interest for wheat breed- 2016.​03.​006
ing. Genet Resour Crop Evol 70:235–250. https://​doi.​org/​ Roldan-Ruiz I, Dendauw J, Van Bockstaele E, Depicker A, De
10.​1007/​s10722-​022-​01429-z Loose M (2000) AFLP markers reveal high polymorphic
Padilla G, Cartea ME, Velasco P, de Haro A, Ordás A (2007) rates in ryegrasses (Lolium spp.). Mol Breed 6:125–134.
Variation of glucosinolates in vegetable crops of Bras- https://​doi.​org/​10.​1023/A:​10096​80614​564
sica rapa. Phytochemistry 68:536–545. https://​doi.​org/​10.​ Šamec D, Pavlović I, Radojčić Redovniković I, Salopek-Sondi
1016/j.​phyto​chem.​2006.​11.​017 B (2018) Comparative analysis of phytochemicals and
Pathirana I, Thavarajah P, Siva N, Wickramasinghe AN, Smith activity of endogenous enzymes associated with their
P (2017) Moisture deficit effects on kale (Brassica olera- stability, bioavailability and food quality in five Brassi-
cea L. var. acephala) biomass, mineral, and low molecular caceae sprouts. Food Chem 269:96–102. https://​doi.​org/​
weight carbohydrate concentrations. Sci Hortic 226:216– 10.​1016/j.​foodc​hem.​2018.​06.​133
222. https://​doi.​org/​10.​1016/j.​scien​ta.​2017.​08.​050 Šamec D, Kruk V, Ivanišević P (2019a) Influence of seed ori-
Peakall ROD, Smouse PE (2006) GENALEX 6: genetic analy- gin on morphological characteristics and phytochemicals
sis in Excel. Population genetic software for teaching and levels in Brassica oleracea var. acephala. Agronomy
research. Mol Ecol Notes 6:288–295. https://​doi.​org/​10.​ 9:502. https://​doi.​org/​10.​3390/​agron​omy90​90502
1111/j.​1471-​8286.​2005.​01155.x Šamec D, Urlić B, Salopek-Sondi B (2019b) Kale (Brassica
Petit J, Salentijn EMJ, Paulo MJ, Thouminot C, van Dinter BJ, oleracea var. acephala) as a superfood: Review of the sci-
Magagnini G, Gusovius HJ, Tang K, Amaducci S, Wang entific evidence behind the statement. Crit Rev Food Sci
S, Uhrlaub B, Müssig J, Trindade LM (2020b) Genetic Nutr 59:2411–2422. https://​doi.​org/​10.​1080/​10408​398.​
variability of morphological, flowering, and biomass qual- 2018.​14544​00
ity traits in hemp (Cannabis sativa L.). Front Plant Sci Schuelke M (2000) An economic method for the fluorescent
11:102. https://​doi.​org/​10.​3389/​fpls.​2020.​00102 labeling of PCR fragments. Nat Biotechnol 18:233–234.
Pipan B, Šuštar-Vozlič J, Meglič V (2011) Cultivation, varietal https://​doi.​org/​10.​1038/​72708
structure and possibilities for cross-pollination of Brassica Sebastien L, Julie J, Francois H (2008) FactoMineR: an R
napus L. in Slovenia. Acta Agric Slov 97:247. https://​doi.​ package for multivariate analysis. J Stat Softw 25(1):1–18.
org/​10.​2478/​v10014-​011-​0019-3 https://​doi.​org/​10.​18637/​jss.​v025.​i01
Pipan B, Šuštar-Vozlič J, Meglič V (2013) Genetic differen- Shannon CE, Weaver W (1949) The mathematical theory of
tiation among sexually compatible relatives of Brassica communication. Bell Sys Tech J 27:379–423. https://​doi.​
napus L. Genetika 45:309–327. https://​doi.​org/​10.​2298/​ org/​10.​1002/j.​1538-​7305.​1948.​tb013​38.x
GENSR​13023​09P Sharma D, Nanjundan J, Singh L, Singh SP, Parmar N, Sujith
Pipan B, Žnidarčič D, Meglič V (2017) Evaluation of genetic Kumar MS, Singh KH, Mishra AK, Singh R, Verma KS
diversity of sweet potato [Ipomoea batatas (L.) Lam.] on (2020) Genetic diversity in leafy mustard (Brassica jun-
different ploidy levels applying two capillary platforms. cea var. rugosa) as revealed by agro-morphological traits
J Hortic Sci Biotechnol 92:192–198. https://​doi.​org/​10.​ and SSR markers. Physiol Mol Biol Plants 26:2005–2018.
1080/​14620​316.​2016.​12499​63 https://​doi.​org/​10.​1007/​s12298-​020-​00883-2

Vol:. (1234567890)
13
Genet Resour Crop Evol (2024) 71:1221–1239 1239

Shen Y, Wang J, Shaw RK, Yu H, Sheng X, Zhao Z, Li S, Gu Thavarajah P, Abare A, Basnagala S, Lacher C, Smith P,
H (2021) Development of GBTS and KASP panels for Combs GF Jr (2016) Mineral micronutrient and prebi-
genetic diversity, population structure, and fingerprinting otic carbohydrate profiles of USA-grown kale (Brassica
of a large collection of broccoli (Brassica oleracea L. var. oleracea L. var. acephala). J Food Compos Anal 52:9–15.
italica) in China. Front Plant Sci 12:655254. https://​doi.​ https://​doi.​org/​10.​1016/j.​jfca.​2016.​07.​003
org/​10.​3389/​fpls.​2021.​655254 Tonguç M, Griffiths PD (2004) Genetic relationships of Bras-
Singh SR, Ahamed N, Kumar D, Srivatsava KK, Yousuf S, Mir sica vegetables determined using database derived simple
A (2017) Genetic divergence assessment in kale (Brassica sequence repeats. Euphytica 137:193–201. https://​doi.​org/​
oleracea L. var. acephala (DC.) Alef) by using the multi- 10.​1023/B:​EUPH.​00000​41577.​84388.​43
variate analysis. J Hortic Sci 12:42–48 Tortosa M, Velasco P, Afonso D, Padilla G, Ríos D, Soengas
Soengas P, Cartea ME, Francisco M, Lema M, Velasco P P (2017) Characterization of a Spanish Brassica oleracea
(2011) Genetic structure and diversity of a collection of collection by using molecular and biochemical markers.
Brassica rapa subsp. rapa L. revealed by simple sequence Sci Hortic 219:344–350. https://​doi.​org/​10.​1016/j.​scien​ta.​
repeat markers. J Agric Sci 149:617–624. https://​doi.​org/​ 2017.​03.​021
10.​1017/​S0021​85961​10001​3X Villanueva RAM, Chen ZJ (2019) ggplot2: elegant graphics for
Soriano JM, Villegas D, Aranzana MJ, García del Moral LF, data analysis
Royo C (2016) Genetic structure of modern durum wheat Wei T, Simko V, Levy M, Xie Y, Jin Y, Zemla J (2017) Pack-
cultivars and Mediterranean landraces matches with their age ‘corrplot.’ Statistician 56:e24
agronomic performance. PLoS ONE 11:e0160983. https://​ Yimer O, Mohammed W, Teju E (2021) Correlation and path
doi.​org/​10.​1371/​journ​al.​pone.​01609​83 coefficient analysis studies in Ethiopian Mustard (Bras-
Šutković J, Glamočija P, Karić L, Yildirim A (2021) Genetic sica carinata A. Braun). Afr J Plant Sci 15:250–256.
characterization of Brassica oleracea L. var. acephala https://​doi.​org/​10.​5897/​AJPS2​021.​2156
DC cultivars from Bosnia and Herzegovina. Genet Appl. Zhu S, Niu E, Shi A, Mou B (2019) Genetic diversity analysis
5:40–50. https://​doi.​org/​10.​31383/​ga.​vol5i​ss1pp​40-​50 of olive germplasm (Olea europaea L.) with genotyping-
Takezaki N, Nei M, Tamura K (2010) POPTREE2: Software by-sequencing technology. Front Genet 10:755. https://​
for constructing population trees from allele frequency doi.​org/​10.​3389/​fgene.​2019.​00755
data and computing other population statistics with Win-
dows interface. Mol Biol Evol 27:747–752. https://​doi.​ Publisher’s Note Springer Nature remains neutral with regard
org/​10.​1093/​molbev/​msp312 to jurisdictional claims in published maps and institutional
Terlević A, Temunović M, Bogdanović S, Grgurev M, Ljubičić affiliations.
I, Rešetnik I (2023) Morphological and environmental
variability of Dianthus sylvestris (Caryophyllaceae) in the
Balkan Peninsula. Bot J Linn Soc 201:377–389. https://​
doi.​org/​10.​1093/​botli​nnean/​boac0​58

Vol.: (0123456789)
13