Maize (Zea mays L. 2n = 20) is the world’s most widely distributed crop and one of the major and popular cereals grown worldwide (Shiferaw et al. 2011). Maize is also considered to be one of the most diverse crops, with great diversity in morphological and physiological traits and extensive polymorphism in its DNA sequences (Thakur et al. 2017). Therefore, these diversity traits are being used as tools for maize breeding programs (Tanavar et al. 2014). In South Sudan, maize is the second most cultivated cereal after sorghum, occupying an estimated 207,714 hectares, which is about 21% of the country’s total cereal area (FAO 2022). Although maize is considered as one of the most important cereals that make up the main source of food for local populations in many states, South Sudan still relies on imports from neighboring countries for most of its needs for maize. Out of 870 metric tons of domestic annual cereal production, maize share 100 metric tons per year (FAO 2022), which is insufficient to meet local con-sump-tion leading the country to food insecurity. In particular, despite the tropical savanna climate of South Sudan, maize is cultivated in all agro-ecological regions. However, climate variation in South Sudan sometimes has negative effects on maize cultivation, especially in semiarid areas with low rainfall (mostly far northern areas) and areas affected by high floods as a result of heavy rain and poor water drainage (mostly central and south-eastern areas), which contributes to reductions in the yields from small-scale farms (AfDB 2019). Such climate differences influence traditional farmers by encouraging them to maintain their own landrace varieties, which suit the growing conditions in each region (FAO 2022). Hence, serious steps are needed to improve local production of maize by adopting breeding strategies such as hybrid breeding and improved open pollinated varieties to maximize productivity in order to satisfy local demand. Therefore, exploring the available genetic resources is critical for upgrading maize yield efficiency as well as providing breeders with insights to select superior germplasm for developing varieties with desirable traits. Consequently, determining the level of variation of quantitative traits available in genetic resources is useful for breeding programs because such traits reflect the agronomic use of germplasm (Darrudi et al. 2018).

Morphological differences in crop germplasm between regions of cultivation area have attracted considerable interest in genetic studies of landrace varieties in each region. Phenotypic variation has been utilized to characterize and manage the genetic diversity for various crop species because it is a strong determinant of agronomic value as well as revealing differences between cultivars or varieties (Cömertpay et al. 2012). Although one of the limitations of morphological characteristics is that they are affected by environmental influences (Belalia et al. 2019), they allow simple grouping of landrace accessions and also help in identifying specific germplasm for breeding (Rohini et al. 2020). Multivariate analysis is a popular statistical instrument for morphological variation studies, and it is considered a useful approach for genetic variation estima-tion, especially as it allows many traits to be estimated jointly (Barth et al. 2022). Among multivariate techniques, principal component analysis (PCA) is one of the preferred methods and is widely used for morphological characteri-zation of populations (Malik et al. 2014; Khavari et al. 2014). PCA allows multicollinear data to be evaluated and the traits most suitable for classification to be determined. PCA has been employed successfully for phenotypic characterization of maize as well as for various other crop species (Darrudi et al. 2018; Sa et al. 2019; Ha et al. 2021). Utilization of this method will help in forming an accurate estimation of the level of phenotypic variation in maize accessions collected from South Sudan. In a previous study, we used simple sequence repeat (SSR) or micro-satellite markers for genetic diversity studies of maize landrace accessions collected from South Sudan, and provided the first report about the genetic variation of maize landrace accessions of South Sudan (Mathiang et al. 2022). Recently, many tools have become available for studying the genetic relationships between landrace germplasm using various molecular markers; however, morphological characterization is the first step in germ-plasm evaluation and classification (Smith et al. 1990). Even today, morphological data continues to be used as a useful analysis method for the management and preser-vation of collected genetic resources. Therefore, combining information on both phenotypic and molecular variation will enable complete and comprehensive knowledge to be formed accurately about the level of genetic diversity present in the maize landrace accessions of South Sudan.

For this study, we collected 31 maize landrace acces-sions from local farmers’ field in various climatic zones of South Sudan, aiming to investigate the phenotypic vari-ations among them. The results of current study will provide maize breeders with information on the morpho-logical differentiation among maize landrace accessions which will aid in the development of improved maize varieties.

Plant materials

The materials used for this study were 31 maize landrace accessions collected from the fields of different farmers across South Sudan (17 accessions from the southern region, 7 accessions from the central region, and 7 accessions from the northern region). These maize landrace accessions were divided into 3 populations based on their geographi-cal distribution, namely northern, central, and southern South Sudan. Information about the 31 maize landrace accessions collected from South Sudan is presented in Table 1 and Fig. 1.

Table 1 . Summary of 31 maize landrace accessions of South Sudan used for morphological analysis.

Regions	No.	Abbr.	State, County, Payam or Village	Country	Type
Southern accession	1	BO1	Gonglei, Bor	South Sudan	Landrace
	2	BO2	Gonglei, Bor	South Sudan	Landrace
	3	BO3	Gonglei, Bor	South Sudan	Landrace
	4	GO1	Central Equatoria, Gondokoro	South Sudan	Landrace
	5	GO2	Central Equatoria, Gondokoro	South Sudan	Landrace
	6	GO3	Central Equatoria, Gondokoro	South Sudan	Landrace
	7	GO4	Central Equatoria, Gondokoro	South Sudan	Landrace
	8	YA	Western Equatoria, Yambio	South Sudan	Landrace
	9	RJ1	Central Equatoria, Rajaf county	South Sudan	Landrace
	10	RJ2	Central Equatoria, Rajaf county	South Sudan	Landrace
	11	RJ3	Central Equatoria, Rajaf county	South Sudan	Landrace
	12	RJ4	Central Equatoria, Rajaf county	South Sudan	Landrace
	13	TO	Torit, Eastern Equatoria	South Sudan	Landrace
	14	MN1	Eastern Equatoria, Mangala	South Sudan	Landrace
	15	MN2	Eastern Equatoria, Mangala	South Sudan	Landrace
	16	MN3	Eastern Equatoria, Mangala	South Sudan	Landrace
	17	MA	Eastern Equatoria, Magwi	South Sudan	Landrace
Central accession	18	MD1	Lakes State, Madhok	South Sudan	Landrace
	19	MD2	Lakes State, Malok	South Sudan	Landrace
	20	AR	Lakes State, Karic	South Sudan	Landrace
	21	AD1	Lakes State, Adull	South Sudan	Landrace
	22	AD2	Lakes State, Kolniet	South Sudan	Landrace
	23	KO	Warrap State, Kuajok	South Sudan	Landrace
	24	TO	Warrap State, Tonj	South Sudan	Landrace
Northern accession	25	RE	Upper Nile, Renk	South Sudan	Landrace
	26	BE	Unity State, Bentiu	South Sudan	Landrace
	27	AW	Northern bahr el-ghazal, Awil	South Sudan	Landrace
	28	YD1	Ruweng Administrative Area, Yida	South Sudan	Landrace
	29	YD2	Ruweng Admistrative Area, Yida	South Sudan	Landrace
	30	RA	Western bahr el-ghazal, Raja	South Sudan	Landrace
	31	WA	Western bahr el-ghazal, Wau	South Sudan	Landrace

Figure 1. Map of South Sudan showing collection sites of 31 maize landrace accessions (refer to Table 1 for accession numbers and remainder of the summary).

Morphological characteristic survey

In our study, to evaluate the morphological variation of the 31 maize landrace accessions collected from South Sudan, 10 individuals of each accession were cultivated from the middle of May to the end of October in 2022 at the research field of Kangwon National University, Chuncheon, Gangwon-do. We examined 7 quantitative characteristics: days to tasseling (DT), days to silking (DS), plant height (PH), ear height (EH), stem width (SW), leaf width (LW), and leaf length (LL) for each maize accession (Table 2).

Table 2 . Morphological characteristics used in morphological variation analysis of maize landrace accessions.

Code	Morphological characters	Category	Time of detecting	Methods of detecting
QN1	Days of tasseling (DT)	days	Flowering stage	Days from seedling to tasseling
QN2	Days of silking (DS)	days	Flowering stage	Days from seedling to silking
QN3	Plant height (PH)	cm	Flowering stage	Length from the ground surface to under corn tassel
QN4	Ear height (EH)	cm	Flowering stage	Length from the surface to the second kernel
QN5	Stem width (SW)	cm	Flowering stage	Stem width near the kernel
QN6	Leaf length (LL)	cm	Flowering stage	Length of the leave near the kernel
QN7	Leaf width (LW)	cm	Flowering stage	Leafe width near the kernel

Data analysis

PCA was performed to detect any morphological dif-ferences between and within the maize accessions col-lected from the different regions of South Sudan. To analyze the morphological variation of the 31 maize land-race accessions of South Sudan, multivariate analysis was performed using the Microsoft Excel Statistical Analysis System Program and NTSYS-pc V2.1 Program (Rohlf 1998). SPSS software was used to perform correlation analysis for the 7 quantitative characteristics of the 31 maize landrace accessions of South Sudan.

Morphological variation among maize landrace accessions.

All maize landrace accessions collected from South Sudan used in this study showed considerable morphological differences in accordance with their collection area (southern, central, and northern regions). The morphological characteristics of the 31 maize landrace accessions for the 7 agricultural characteristics are shown in Supplementary Table S1.

To understand the differences in morphological variation among the maize accessions of the 3 regions (southern region accessions, central regions accessions, and northern region accessions) of South Sudan, the mean, standard deviation, and minimum and maximum values for all characteristics were calculated for accessions of the 3 regions (Table 3). Based on the survey of the 7 quantitative characteristics for accessions of the 3 regions of South Sudan, the average of DT for the maize landrace accessions of southern, central, and northern regions was 83.0 ± 6.24 (76-97), 91.9 ± 9.06 (80-99), and 83.6 ± 4.22 (79-89), respectively. The average of DS was 86.1 ± 6.89 (78-101), 93.1 ± 10.3 (79-101), and 82.3 ± 4.26 (78-90) for the accessions of southern, central, and northern regions, respectively. The average values of PH were 253.7 ± 34.7 (210-350), 252.6 ± 19.8 (214-273), and 266.6 ± 12.8 (250-281) cm for the accessions of southern, central, and northern regions, respectively. The average values of EH for the maize landrace accessions of southern, central, and northern regions were 137.6 ± 30.9 (96-225), 141.3 ± 21.0 (103-170), and 156.5 ± 16.2 (134-176) cm, respectively. The average values of SW for the maize landrace acces-sions of southern, central, and northern regions were 3.0 ± 0.86 (2.1-5.4), 3.5 ± 0.73 (2.5-4.4), and 3.8 ± 0.85 (2.8-5.4) cm, respectively. The average values of LL were 98.6 ± 12.6 (66.5-120), 96.3 ± 4.85 (89.3-103), and 107.1 ± 12.9 (94.3-131) cm for the maize landrace accessions of southern, central, and northern regions, respectively. The average values of LW were 9.7 ± 1.08 (7.3-11.7), 10.2 ± 1.64 (7.2-11.8), and 10.5 ± 1.06 (8.5-12) cm for the accessions of southern, central, and northern regions, respectively (Table 3).

Table 3 . Morphological characteristics of 7 quantitative characteristics among 31 maize landrace accessions of 3 regions of South Sudan.

Code	Morphological Characters	Region	Category/Unit	Mean ± SD	Min			Max
					Values	Genotype		Values	Genotype
QN1	Days of tasseling (DT)	SA	Days	83.0 ± 6.24	76	MN1		97	GO1
		CA	-	91.9 ± 9.06	80	AD2		99	AR
		NA	-	83.6 ± 4.22	79	AW		89	AW
QN2	Days of silking (DS)	SA	Days	86.1 ± 6.89	78	MN1		101	GO1
		CA	-	93.1 ± 10.3	79	AD2		101	AR
		NA	-	82.3 ± 4.26	78	WA		90	RE
QN3	Plant height (PH)	SA	cm	253.7 ± 34.7	210	GO3		350	TR
		CA	-	252.6 ± 19.8	214	MD1		273	AR
		NA	-	266.6 ± 12.8	250	RE		281	BE
QN4	Ear height (EH)	SA	cm	137.6 ± 30.9	96	GO3		225	TR
		CA	-	141.3 ± 21.0	103	MD1		170	KO
		NA	-	156.5 ± 16.2	134	WA		176	YD2
QN5	Stem width (SW)	SA	cm	3.0 ± 0.86	2.1	GO4		5.4	RJ2
		CA	-	3.5 ± 0.73	2.5	MD1		4.4	AD1
		NA	-	3.8 ± 0.85	2.8	RE		5.4	YD1
QN6	Leaf length (LL)	SA	cm	98.6 ± 12.6	66.5	BO1		120	TR
		CA	-	96.3 ± 4.85	89.3	AD2		103	TO
		NA	-	107.1 ± 12.9	94.3	YD1		131	WA
QN7	Leaf width (LW)	SA	cm	9.7 ± 1.08	7.3	RJ3		11.7	GO1
		CA	-	10.2 ± 1.64	7.2	AD2		11.8	KO
		NA	-	10.5 ± 1.06	8.5	RE		12	AW

SA: southern region accessions, CA: central region accessions, NA: northern region accessions.

Meanwhile, we performed a correlation analysis to detect the morphological relationships between the 7 agronomic characteristics in the 31 maize landrace acces-sions collected from the 3 regions of South Sudan (Table 4). In our study, the 7 agronomic characteristics showed statistically significant positive or negative correlation coefficients at significance levels of 0.05 and 0.01. Among all the morphological characteristics, the combinations between DT (QN1) and DS (QN2) (0.960**), between PH (QN3) and EH (QN4) (0.767**), between QN3 and SW (QN5) (0.518**), between QN4 and QN5 (0.564**), and between QN4 and LL (QN6) (0.496**) showed compara-tively higher positive correlation coefficients at signifi-cance levels of 0.01. Further, the combination between QN3 and QN6 (0.373*), between QN3 and LW (QN7) (0.440*), between QN4 and QN7 (0.401*), and between QN5 and QN6 (0.403*) showed comparatively higher positive correlation coefficients at the 0.05 significance level. The remaining characteristics showed low correla-tions in both positive and negative directions (Table 4).

Table 4 . Pearson correlation coefficient for 7 quantitative characteristics in 31 maize landrace accessions of South Sudan.

Morphological characters	QN1	QN2	QN3	QN4	QN5	QN6	QN7
QN1		0.960**	0.120	0.044	‒0.09	0.119	0.036
QN2			0.127	0.015	‒0.166	0.091	‒0.003
QN3				0.767**	0.518**	0.373*	0.440*
QN4					0.564**	0.496**	0.401*
QN5						0.403*	0.324
QN6							0.238

**Significance at P < 0.01, *Significance at P < 0.05.

Principle component analysis for maize landrace accessions

PCA in this study was performed to evaluate the mor-phological differentiation among the 31 maize landrace accessions. The first principal component (PC1) and second principal component (PC2) accounted for 40.9% and 28.6%, respectively, of the total variance (Table 5). Among all morphological characteristics, 3 characteristics, PH (QN3), EH (QN4), and SW (QN5) made a significant contribution to the positive direction of PC1. Also DT (QN1) and DS (QN2) made a remarkable contribution in the positive direction on PC2 (Table 5). Along axis 1 of the PCA (Fig. 2), most maize landrace accessions of South Sudan were not clearly segregated based on their collection areas of the southern, central, and northern regions. In detail, except for several exceptional accessions, most maize landrace acces-sions of the southern region were formed in a group on the negative side on the first axis, and most maize landrace accessions of the central region were formed in a group on the positive and negative sides on the first axis. Also, most maize landrace accessions of the northern region, with a few exceptional accessions, were formed in a group on the positive side on the first axis. 

Table 5 . Cumulative variances of first and second principal components and loadings of 7 quantitative cha-racteristics on each principal component.

Code	Morphological characteristics	Eigenvectors
		1	2
QN4	Ear height (EH)	0.88	‒0.077
QN3	Plant height (PH)	0.855	0.03
QN5	Stem width (SW)	0.732	‒0.278
QN6	Leaf length (LL)	0.654	0.051
QN7	Leaf width (LW)	0.603	0.051
QN1	Days of tasseling (DT)	0.149	0.971
QN2	Days of silking (DS)	0.108	0.984
Cumulative variance (%)		40.9	28.6

Figure 2. Scatter diagram of 31 maize landrace accessions based on first and second principal components. ◇: Southern region accessions, ▲: Central region accessions, ●: Northern region accessions.

Morphological characterization is a fundamental method for estimating the variation and genetic diversity in plant breeding programs (Park et al. 2015; Vathana et al. 2019). Although it presents many limitations such as low polymorphism and environmental influence on phenotypic expression, it provides critical information that is necessary for characterizing genetic resources (Belalia et al. 2019). The morphological trait phenotypes of crop species are important characteristics that appear when crops adapt to their environment, and the phenotypic plasticity of crop species may be related to the specific environmental conditions in which crops live (Schlichting 1986). Thus, environmental conditions play an important role in the morphological variation in crop species. Also, morphologi-cal variation in crop species can evolve in response to changing environments and can show different patterns across cultivation regions because of geographical distri-bution.

Descriptive statistics used in our study were calculated to account for morphological variation in maize landrace accessions collected from South Sudan. The data obtained provides measurements and forms the basis for almost all quantitative analysis of data, including the mean, mini-mum, maximum, and standard deviation of each quanti-tative characteristic in the accessions of maize germplasm (Table 3). Previously there have been few reports on the results of cultivation tests in domestic cultivation fields for foreign maize germplasm accessions from areas such as Africa including South Sudan. Thus, in the present study, morphological characteristics were investigated to under-stand the morphological variation of 31 maize landrace accessions collected from farmers' fields in South Sudan. Even though the growth survey experiment was conducted in Korean fields, an examination of the morphological characteristics of the 31 maize landrace accessions col-lected in the southern, central, and northern regions of South Sudan showed that they had different growth characteristics depending on the collection region. In our study, although the maximum and minimum values for each accession collected from the 3 regions of South Sudan did not show any particular tendency based on the collection region, for the average value of the accessions of each of the 3 regions, the accessions of CA region showed the highest values for DT (QN1) and DS (QN2), while the accessions of SA region showed the highest values for PH (QN3), EH (QN4), SW (QN5), and LL (QN6). Despite the results being from an investigation in the climatic environment of South Korea, the maize landrace acces-sions collected in the central and southern regions of South Sudan showed comparatively good growth characteristics. In addition, relationships between morphological characte-ristics were examined using correlation coefficients, and the results of a simple correlation coefficient analysis revealed generally the existence of both significant positive and negative correlations between the 7 morphological characteristics. However, among the combinations, the combinations between QN1 and QN2 (0.960**), between QN3 and QN4 (0.767**), QN3 and QN5 (0.518**), and between QN4 and QN5 (0.564**), and QN4 and QN6 (0.496**) showed significant high positive correlation coefficient compared with the other characteristics at P < 0.01 (Table 4). Similar results for a high and positive correlation between agronomic traits were recorded in several other reports, especially for maize (Cömertpay et al. 2012; Iqbal et al. 2015; Belalia et al. 2019). This positive correlation between traits indicates that improve-ment of any characteristic may have a positive effect on desired traits, as a correlation between traits is caused by the presence of linked genes and the occurrence of an inter allelic effect between genes (Cömertpay et al. 2012).

PCA helps in determining the most relevant traits that can be employed as indicators by explaining much of the overall variation in the original set of variables, and the traits that contribute greatly to the divergence provide considerable affirmation for deciding on a cluster for the purpose of selection of parents for hybridization (Thakur et al. 2017). Therefore, PCA was carried out to classify the maize landrace accessions based on the most discrimina-ting characteristics, and the results disclosed that PC1 and PC2 accounted for 40.9% and 28.6%, respectively, of the total variance. Among the evaluated characteristics, PH (QN3), EH (QN4), SW (QN5), LL (QN6), and LW (QN7) contributed positively for PC1; while DT (QN1) and DS (QN2) had a large positive contribution for PC2. The PCA based grouping for maize landrace accessions is in good agreement with previous reports by Hartings et al. (2008), Mijangos-Cortés et al. (2007), Jarić et al. (2010), and Belalia et al. (2019). These studies reported that plant and ear height characteristics were the most discriminative characteristics for identifying maize landrace populations in Italy, Mexico, Eastern Serbia, and Algeria, respectively. Therefore, these characteristics, which revealed a maximum variance with relatively high vectors, are important cha-racteristics for distinguishing phenotypic variation in maize landrace accessions of South Sudan. Therefore, selection with regard to these characteristics should be taken into consideration when choosing parent lines to ensure yield improvement.

Scatter plot based on PC1 and PC2 showed the existence of 3 major groups (Fig. 2). Group 1 allocated in the first quadrant on the positive side on the first axis included 5 maize accessions that had higher values for DT (QN1), DS (QN2), and PH (QN3). Group 2 allocated in the second and third quadrants on the positive side on the first axis comprised 14 maize accessions that showed low values for PH (QN3) and EH (QN4). Group 3 allocated in the fourth quadrant on the negative side on the first axis included 10 maize accessions that showed high values for PH (QN3) and EH (QN4). PH (QN3), EH (QN4), and DS (QN2) are important characteristics associated with flowering time, which is a crucial characteristic that contributes to crop yield (Zhang et al. 2011; Vathana et al. 2019); therefore, this result provides useful details for choosing lines for the improvement of characteristics associated with flowering time. In general, the PCA did not clearly group the different accessions based on their areas of origin, which corres-ponds to the findings of Hartings et al. (2008), Sharma et al. (2010), and Belalia et al. (2019). These studies investigated the phenotypic traits of maize accessions from Italy, the north-eastern Himalayas, and Algeria respectively and found that the accessions were not clustered together based on their geographical region. Mixing of germplasm accessions among local farms could be the reason behind the weak relationship among the accessions from the same region, which is the same finding that was reported in our previous report for accession clustering based on microsatellite marker data (Mathiang et al. 2022).

In conclusion, despite the morphological characterization studies being conducted in the Korean climate, our study sheds light on and provides details about morphological variation among maize landrace accessions collected from South Sudan. The characteristics that show high variability among the accessions are the most discriminating cha-racteristics and were the best for describing the morpho-logical variation for our explored genotypes. The grouping results based on PCA offer information for selecting suitable genetic materials for breeding. This study together with our previous report provides a comprehensive under-standing of genetic diversity present in South Sudan maize, which had been little investigated. These findings will be crucial in future for advancing molecular breeding studies such as marker-assisted selection (MAS), genome wide association study (GWAS), and genome selection. 

This study was supported by the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (NRF-2021R1A6A1 A03044242) and the Golden Seed Project (No. 213009- 05-1-WT821, PJ012650012017), Ministry of Agriculture, Food, and Rural Affairs (MAFRA), Ministry of Oceans and Fisheries (MOF), Korea Forest Service (KFS), Republic of Korea.