Crop diversification in the context of climate change is now a necessary response to food insecurity. Increasingly adverse weather conditions, such as droughts and high temperatures, are affecting several countries in the sub-Saharan zone. As a result, these countries need to reduce agricultural losses by diversifying the food basket with a wide range of under-utilized crop species adapted to marginal environments. These under-utilized crops can thrive in the stressful, low-input growing conditions that limit agricultural productivity worldwide and will become more prevalent with climate change. This is the example of amaranth, a vegetable crop grown in many countries around the world (Dinssa et al. 2016; Kahane et al. 2005). It is highly adaptable to all agroecological conditions and grows rapidly (National Research Council 2006; Westphal et al. 1985). In West Africa, it is one of the most popular leafy vegetables on many vegetable farms, especially in and around large towns (Tchiadje 2008). Amaranth belongs to the Amaranthaceae family and is a C4 dicotyledonous plant (Clouse et al. 2016; Thapa and Blair 2018). It comprises around 60-70 species grouped into three subgenera: Amaranthus Albersi, Amaranthus Amaranthus and Amaranthus Acnida. Some species are cultivated as leafy vegetables and nutritious pseudocereals, others are cultivated ornamentals and the rest are considered weeds (Das 2016; Sauer 1967). In Burkina Faso, many amaranth species such as Amaranthus cruentus, Amaranthus hypochondriacus and Amaranthus dibius are grown as leafy vegetables, the best known and most widely grown being Amaranthus cruentus (Achigan-Dako et al. 2014; Ouedraogo et al. 2023). These species have a high socio-economic and nutritional value for local populations (Ouedraogo et al. 2023). Consequently, knowledge and management of their genetic resources are key steps in achieving this objective. Phenotypic markers are of great value in germplasm evaluation studies (Bretting and Wildrlechner 1995) and also for revealing differences between varieties (Gilliland et al. 2000). Previous studies in Burkina Faso have shown great morphological variability within cultivated species and have identified seven morphotypes whose main classification criteria have been the color and shape of leaves and inflorescence (Ouedraogo et al. 2021). As estimates of genetic variability based on morphological traits have the disadvantage of being influenced by both environmental and genetic factors, morphological analysis may not provide a correct estimate of genetic diversity. In this sense, a global approach including molecular markers is needed to analyse diversity and support the conservation, management and development of plant genetic resources (Hammer and Teklu 2008). Thus, in recent years, considerable emphasis has been placed on the development and use of molecular markers in all major crops (Hirata et al. 2006; Khaing et al. 2013). PCR-based markers have been successfully developed such as RFLPS, RAPD, and SSR (Lee et al. 2008; Wassom and Tranel 2005). Among these markers, SSR have become the markers of choice due to several desirable attributes, including their abundance, multi-allelic and codominant nature, high level of reproducibility and transferability between related species (Gupta and Varshney 2000; Lahaye et al. 2008; Purty and Chatterjee 2016). Several of these SSR markers have been successfully exploited by several authors (Costea et al. 2006, Kaing et al. 2013) on species of the genus Amaranthus for genetic diversity analysis. Our study aims to test 11 SSR markers on a collection of 72 accessions of amaranth cultivated in Burkina Faso. The aim is to assess intraspecific variability and interspecific variability and to establish the level and structure of genetic diversity in Amaranthus germplasm collected in the three climatic zones of Burkina Faso.

Plant material

A total of 72 amaranth accessions were collected in Burkina Faso (Table 1). These included 58 accessions of the green morphotype, two accessions of the purple morphotype, two accessions of the light green morphotype with purple base, six accessions of the green-purple morphotype, two accessions of the purple morphotype and 2 accessions of the dark green morphotype. The accessions were collected from growers at market garden sites in different climatic zones of Burkina Faso. A total of 22 accessions were collected in three provinces of the Sahelian zone, 44 accessions in the Sudano-Sahelian zone and 6 accessions in the province of Houet in the Sudanian zone. The morphological characteristics of the accessions and their place of origin are given in Table 1 below.

Microsatellite markers

A set of 11 polymorphic microsatellite markers developed by Mallory et al. (2008) and Lee et al. (2008) were used in this study (Table 2). The loci were hybridized at temperatures between 41.4°C and 62°C.

DNA isolation

DNA was extracted from young leaves using a combination of DNA extraction protocols described by Saghai-Maroof et al. (1984) and Agbangla et al. (2002). Leaves from each sample (0.2 g) were ground in 1250 ul of Tris-EDTA-Sorbitol buffer and centrifuged at 10,000 rpm for 10 min at 4°C. After this step, the supernatant was removed and 750 µl of MATAB (Mixed Alkyl Trimethyl Ammonium Bromide) buffer pre-warmed to 65°C was added to the tubes and incubated at 65°C for 2 h 30 min. At the end of the incubation 750 ul of a solution of chloroform: isoamyl alcohol (24:1) were added at room temperature and shaken by inverting tubes for 5 min, centrifuged at 12,000 rpm for 10 min at 4°C to accelerate the separation phase. The aqueous phase was recovered and an equal volume of isopropanol stored at 4°C was added and kept at -20°C for at least one hour to precipitate the DNA before centrifuging again at 12,000 rpm for 10 min at 4°C. At the end of centrifugation, the DNA pellet was collected, washed with 70% ethanol and then dried at room temperature for 30 min before dissolving in 500 ml Tris-EDTA buffer and storing at -20°. The quality and integrity of the isolated DNA was checked by 1% agarose gel electrophoresis.

PCR amplification

PCR reactions were performed in a final volume of 20 µl containing 1 µl of 3’ primer (forward primer), 1 µl of 5’ primer (reverse primer), 9 µl of Milli-Q water, 4 µl of PCR premix consisting of 1U Taq polymerase, 250 µM Tris-HCL, 10 mM KCl, 1.5 mM MgCl2 and 5 µl of genomic DNA (10 ng/µl).

The reaction mixture was then placed in a thermal cycler for PCR amplification. This amplification was carried out according to a program consisting of an initial denaturation phase at 95°C for 5 min, followed by a series of 40 cycles. Each cycle consisted of a denaturation phase at 94°C for 1 min, hybridization at the temperature (°C) of each primer for 30 s and extension at 72°C for 1 min. At the end of the 40 PCR cycles, a final extension at 72°C for 10 min was carried out, followed by cooling to 4°C until deposition on the agarose gel. The amplification products were subjected to electrophoresis on a 4% agarose gel prepared with a 1 X TBE solution. Deposits were made in the presence of a molecular weight marker ranging in size from 50 to 1500 bp and migration was performed at 100 V for 1 h 30 min in 0.5x Tris Borate EDTA (TBE) buffer. At the end of the migration, a 5% ethidium bromide (BET) solution was used as a developer.

Bands were read using a model DI-01-220 trans illuminator with a 10 mega pixel camera. The bands were identified on the basis of their position on the gel.

Data scoring and analysis

Data analysis was performed at two levels: intrapopulation variability and interpopulation variability. Genetic parameters were calculated using PowerMarker v. 3.25 (Liu and Muse 2005) and GenAlex 6.5 (Peakall and Smouse 2012). The dendogram was reconstructed using the Neighbour Joining (N-J) method using Darwin software version 5.0.155 (Perrier and Jacquemoud-Collet 2006). Population structuring based on an admixture model with correlated allele frequencies was carried out using STRUCTURE 2.3.4 software (Pritchard et al. 2000).

Genetic diversity at intra-population level was described by calculating eight parameters. They were estimated for each locus and also for each morphotype and climatic zone. Major allele frequency (MAF),

The polymorphism rate (P), which is the proportion of polymorphic loci. A population is said to be polymorphic for a given locus if the allele frequency of the most frequent allele is less than 0.95.

Average number of alleles per locus, this reflects the allele richness of a population and is calculated using the formula:

Observed heterozygosity (Ho), represents the average of the frequencies of heterozygotes observed at each of the loci studied. It is equal to the number of heterozygous individuals divided by the total number of individuals in the sample.

Expected heterozygosity (He) expresses the probability that two genes drawn at random from a population are different.

The Shannon diversity index gives an idea of specific diversity.

The fixation index measures the difference between the rate of observed heterozygosity and the rate of expected heterozygosity in a population of individuals found at a distance from the Hardy Weinberg equilibrium (Guo and Thompson 1992).

Polymorphism Information Content (PIC) is a parameter that provides an estimate of the discriminatory power of a locus (Botstein et al. 1980).

where p is the relative frequency of the j th pattern for SSR marker i (Botstein et al. 1980).

Genetic diversity (D) between populations, morphotypes, climatic zones and genetic groups was estimated by calculating three parameters.

Nei’s genetic distance (Nei and Chesser 1983) was calculated for the genetic factors studied.

It is used to assess the degree of similarity of their genetic structures, and also to show whether or not groups of plants claimed to belong to different species are part of the same species complex.

Px=∑Xi2, is the probability of identity of the 2 alleles taken at random from the population X

Py=∑yi2, is the probability of identity of the 2 alleles taken at random from the population Y

Pxy=∑XiYi, is the probability of identity of the 2 alleles taken at random, one in X and the other in Y.

Genetic differentiation indexes measure the heterozygote deficit due to differentiation between sub-populations, providing information on the degree of genetic differentiation between sub-populations.

The number of subpopulations (K) was set at 1-10, with 100 replications per K value. The K value was determined by running a series of tests on the sub-populations. The K-value was determined by running a mixture model and correlated allele frequencies. Each run began with 150,000 burn-ins, followed by 200,000 MCMC (Markov chain Monte Carlo) iterations. The ad hoc statistic Delta K was calculated to detect populations using the online program STRUCTURE HARVESTER (Evanno et al. 2005). An individual accession was assigned to a group (subpopulation) if more than 70% of the probability of belonging came from that group.

Genetic diversity and markers polymorphism

The markers tested showed the existence of genetic diversity within the collection, with a polymorphism rate of 100% for each marker. Allele sizes ranged from 50 bp for the AHAAT118 marker to 500 bp for the 71N marker (Table 3). A total of 36 alleles, ranging from two alleles for markers 32N, 78N, 51F and 105N to six alleles for 71N, with an average of 3.27 alleles per marker. The major allele frequency (MAF) ranged from 0.52 for AHAC064 to 0.99 for 78N, with an average of 0.78. The mean value of observed heterozygosity (Ho = 0.26) was lower than that of expected heterozygosity (He = 0.27). Zero observed heterozygosity values (Ho = 00) were recorded for the markers (32N, 78N, 105N and AHAAT38). Expected heterozygosity (He) values greater than 0.50 were observed for markers 71N and AHAC064. A variation in marker po33.lymorphism levels was observed, with PIC values ranging from 2.70 to 51.73% and an average of 22.20%. The fixation index (Fis), which measures the reduction in heterozygosity, ranged from -0.04 to 1, with an average value of 0.33.

Genetic diversity according to morphotype

The genetic diversity parameters calculated vary from one morphotype to another (Table 4). The green morphotype recorded the highest values for genetic parameters compared with the other morphotypes. This morphotype showed a polymorphism rate of 100% with an average of 3.63 alleles and a heterozygosity deficit (Ho = 0.35 and He = 0.37). The other morphotypes showed an excess of heterozygosity. The lowest polymorphism rate was 25% and was observed in accessions of the purple-green morphotype. As for the Shannon diversity index, the values calculated were low and ranged from 0.17 to 0.65 respectively for accessions of the green-purple and green morphotypes. The average fixation index for the different morphotypes was positive (Fis = 0.36).

Inter-morphotype differentiation

The results of genetic differentiation between the six morphotypes are reported in Table 5. The Nei genetic distance calculated between morphotypes varied from 0.017 between the green purple and light green morphotype to 0.347 for the light green and dark green morphotype. The dark green morphotype presented the greatest genetic distance calculated with the five other morphotypes with values varying between 0.259 and 0.347. The light green, purple and green purple morphotypes were the closest genetically with distances varying between 0.017 and 0.037.

The genetic differentiation indices calculated between morphotypes show a strong differentiation between the dark green morphotype and the other morphotypes (0.286 ≤ Fst ≤ 0.452) and between the red morphotype and the four morphotypes (0.157 ≤ Fst ≤ 0.257). In contrast, moderate differentiation was observed between the green, green purple, purple and light green morphotypes (0.05 ≤ Fst ≤0.11).

Genetic diversity according to climatic zones

The results of the genetic differentiation (Table 6) showed that the greatest genetic distance (0.071) was observed between the Sudanian and Sahelian zones and between the Sudanian and Sudano-Sahelian zones (0.081). Moreover, between the Sahelian and Sudano-Sahelian zones, only 4% of the total variability can be attributed to the climate factor. The differentiation index (Fst) varied from 0.044 between accessions from the Sudano-Sahelian zone and those from the Sahelian zone to 0.101 between accessions from the Sudanian zone and those from the Sahelian zone. Thus, genetic differentiation between accessions from different zones was low between those from the Sudano-Sahelian zone and those from the Sahelian zone, moderate between the Sudano-Sahelian and the Sahelian zone, and high between the Sudano-Sahelian zone and the Sahelian zone.

Principal Component Analysis.

The first two axes of the principal coordinate analysis (PCoA) explain respectively around 23.94% and 15.33% of the variation in the genetic distance matrix, i.e. a cumulative 39.27% of the total variance (Fig. 1). The analysis shows a separation of accessions into two subgroups along axis 1. Accessions in subgroup I are more negatively correlated with axis 1 and positively correlated with axis 2, while the majority of accessions in subgroup II are more positively correlated with axis 1 and a few accessions are highly correlated with axis 2. Accessions from the Sudanian zone tended to cluster on the positive side of the first axis, while those from the other two zones were distributed along axis 1 (Fig. 1B).

In terms of morphotypes, there were close and different groupings between accessions (Fig. 1A). Accessions with red, purple, dark green and light green morphotypes clustered closely together, indicating a close genetic relationship. In contrast, the green and purple-green morphotypes showed no distinct clustering patterns, indicating genetic variation in their breasts.

Population structure of amaranth accessions

Analysis revealed that the highest delta K value was observed at K = 2 (Fig. 2). These results indicate that the accessions in the collection can be grouped into two primary composite populations with average individual membership coefficients (Q) greater than or equal to 70% (Fig. 3). Subpopulation 1 comprises 25 accessions collected in the Sudano-Sahelian zone, 7 accessions in the Sahelian zone and one accession in the Sudanian zone. As for subpopulation 2, it is composed of 19 accessions collected in the Sudano-Sahelian zone, 16 in the Sahelian zone and five in the Sudanian zone.

Genetic diversity according to subpopulations

The mean genetic diversity (He) for in sub-population 1 and sub-population 2 of the collection was respectively 0.41 and 0.32 with an average of 0.37 (Table 7). As for the polymorphism information content, the calculated mean values were 0.48 and 0.34 for sub-population 1 and sub-population 2 respectively. Comparative analyses of genetic diversity showed that the accessions of sub-population 1 had a high genetic diversity compared to the accessions of sub-population 2. However, the mean number of alleles in sub-population 2 was higher than in population 1, with an average number of 3.14 for both subpopulations.

Organization of the collection using the Neighbour-Joining method

The organization of the genetic diversity of the amaranth collection in Burkina Faso showed differentiation at the threshold of the 5% confidence interval based on the dissimilarity matrix using the Neighbour-Joining method. This was done independently of the collection sites. However, there is a tendency for groupings to be based on morphotypes (Fig. 4). Group I comprise 34 accessions, including 31 accessions of the green morphotype, two accessions of the dark green morphotype and one accession of the purple morphotype.

Group II, comprising 38 accessions, is subdivided into two subgroups. Subgroup IIa comprises 24 accessions. It is a heterogeneous group containing accessions belonging to the morphotypes identified with the exception of the dark green morphotype. Subgroup IIb comprises 14 accessions, all of the green morphotype.

Permutation tests enabled us to observe genetic proximity between several accessions of the same morphotype and different morphotypes. A degree of similarity of 68% was observed between the green morphotype accession (SAN2) and the purple morphotype accession (SAN3). Significant degrees of similarity were also observed between the two accessions (BOB3 and BOB4) of the light green morphotype (68%), between the two accessions (KAD4 and KAD6) of the dark green morphotype (90%) and between the two accessions (BOB5 and BOB6) of the red morphotype (67%).

Effet des facteurs «zones climatiques» «morphotype» et «groupe génétique» sur la différenciation génétique des accessions de la collection

Analysis of molecular variance (AMOVA) showed that the factors ‘climatic zone’, ‘sub-population’ and ‘genetic group’ were significantly involved in expressing the molecular variability of the collection studied (Table 8). The values of the genetic differentiation index calculated were between 0.05 and 0.10, showing moderate differentiation for these factors. The ‘morphotype’ factor plays a very small role in the expression of variability, with an estimated variance of 0.033 and a contribution to the total variance distribution of 2%. The value of the total genetic differentiation index calculated was low, showing little differentiation between morphotypes. Fixation index values ranged from 28.10% (between climatic zones) to 31.30% (between morphotypes).

The study of the genetic diversity of cultivated species based on a molecular approach, as was the case in this study, is a reliable way of better estimating the diversity of collections of accessions. It also enables genetic resources to be better conserved (Mondini et al. 2009). This study of the genetic diversity of amaranth accessions grown in Burkina Faso showed moderate genetic diversity, in contrast to the high variability observed using morphological markers (Ouedraogo et al. 2021). This difference in variability could be explained in part by the phenotypic plasticity phenomenon, in which the same genotype is capable of producing different phenotypes depending on environmental conditions. Indeed, previous studies have shown that the phenotypic plasticity of amaranths complicates the identification of a morphological marker (Kulakow and Jain, 1990). However, these SSR markers each revealed a polymorphism rate of 100%, reflecting their ability to assess the genetic diversity of amaranth collections. As in the case of the work by Khaing et al. (2013) and Wang and Park (2013), these microsatellite markers make it possible to better assess and understand the level and structure of genetic diversity. In addition, the positive differences between expected and observed heterozygosity associated with the high positive value of the fixation index indicate a high level of inbreeding within the collection studied. This level of inbreeding is one of the reasons of the moderate variability throughout the amaranth collection grown in Burkina Faso. Inbreeding modifies genotypic frequencies, resulting in a loss of genetic variability over generations (Jordana et al. 2003; Mayuri et al. 2019). The high level of inbreeding observed could be essentially due to the way amaranth reproduces. Cross-pollination is a very common phenomenon in amaranth species (Htet and Park 2013; Sauer 1967), and could explain the high variation between accessions and the low level of differentiation between morphotypes identified. This possibility of inter-hybridization could be justified by the ways in which growers manage seeds, such as mixed cropping, seed exchanges or the introduction of other cultivars through the migratory flow. When populations migrate, they take their seeds with them and introduce them into the host localities. As a result, the same morphotypes can be found in several growing areas in different climatic zones (Diouf et al. 2007; Ouedraogo et al. 2023). The reason for moderate genetic diversity could also be due to the size of our collection and the wide collecting area, which covers only one country, unlike previous studies (Khaing et al. 2013; Thapa et al. 2021) whose collections contain a high number of accessions and cover several countries and continents.

The two populations identified by STRUCTURE show that none of the accessions has a genome from just one of the two populations defined. The two populations a priori defined by the model are probably descended from the same ancestor. This structuring model has already been reported by several authors on amaranth collections containing a larger number of species (Kaing et al. 2013; Thapa et al. 2021). As for the genetic tree constructed using the Neighbours Joining method, it shows an organization more related to the accessions and morphotypes identified than to climatic and geographical zone factors. Similar results were found with ISSR markers in a collection of 54 accessions from two climatic zones in Burkina Faso (Ouedraogo et al. 2019). The impossibility of finding a correlation between genetic groups and the geographical origin of accessions could be due to the cosmopolitan nature of the genus and the results of human activities through seed exchanges. Indeed, seed exchange between farmers in Burkina Faso has been observed in previous studies on vegetables including amaranth (Bationo-Kando et al. 2015; Kiébre 2016; Ouedraogo et al. 2023).

Overall, the study shows genetic proximities between most of the morphotypes identified, which could be due to allogamy within the same species or between different species. Another hypothesis is that the environment influences the expression of certain morphological traits. The appearance of new morphological characters from one environment to another could be a means of adapting accessions. Thus, multi-environment studies combined with a molecular study will make it possible to identify the contribution of the environment to the genetic variability of amaranth accessions grown in Burkina Faso. However, this study has provided a basis for better conservation of the genetic resources of this plant in Burkina Faso. It also opens up new prospects for breeding and genetic improvement programs.