Sorghum, [Sorghum bicolor (L.) Moench] is the fifth major cereal grown in the world after corn, rice, wheat and barley. On average, it takes up 42.8 million hectares with a grain production of 63.4 million tons (FAOSTAT, 2012-2016). The major part of cultivated areas (64.3%) is found in Africa, whereas the grain production represents only 42.2% of the world production. The average grain yield for Africa is low (970 kg/ha) compared with that of the other countries in the world (2399 kg/ha). Such differences in yield  between  Africa  and  the  rest  of  the world are due to the technical and economic means used in the production. Bindraban et al. (2000) and Leiser et al. (2013-2014) mentioned that among the major constraints of sorghum production in Africa are the water availability and the deficiency of soil mineral nutrients, particularly in phosphorus. Ayana and Bekele (1998), Mujaju and Chakauya (2008), Ngugi and Maswili (2010), Dossou et al. (2015), Desmae et al. (2016) and,Teshome et al. (2018) observed that the sorghum growing in Africa was still widely based on local varieties that remain an important component of traditional farming systems.

In Burkina Faso, sorghum is the first food crop, followed by millet, corn, rice and fonio. On average it takes up to 42.6% (1.7 million hectares) of cereal growing area with a grain production of 37.5% (1.7 million tons), (MARHASA, 2012-2016). The farming systems of sorghum are still extensive. This is due to the low investment capacity of farmers (low level of agricultural equipment, the non-use or low use of mineral fertilizers, etc.); moreover, sorghum is often cultivated on marginal soils.

Studies have shown that local sorghum varieties were the most widely cultivated in Burkina Faso (Zongo, 1991; Barro-Kondombo et al., 2010; Nebié et al., 2009; Sawadogo et al., 2014). These local varieties mostly (93 to 96%) belong to the guinea botanical race (Harlan and De Wet, 1972). The varieties of this race (, although their low agronomic performance) are preferred by farmers for their rusticity, their sensitiveness to photoperiod which better suit to the production conditions of traditional farming systems; their high stem and grain quality are preferred respectively for various domestic and culinary uses. Some improved sorghum varieties of caudatum botanical race have been developed and released by sorghum research program; despite their higher productivity, they are much less cultivated due to their weak adaptability to the extensive systems production.

The constraints of sorghum growing have always existed in Burkina Faso, but their nature and importance vary according to farming areas. In the North region, sorghum takes up on average 49.6% of cereal growing areas before millet (46.1%). The low agricultural potential of cultivated land (DGAT, 2006), shortening of the rainy season and the bad distribution of rainfall (Paturel et al., 2010), as well as bioagressors attacks, particularly striga (Striga hermonthica) are the major constraints which affect sorghum yields in the region. Some old low performing varieties haven been given up and many new varieties have beenwere introducedin the villages by farmers (Kondombo et al., 2016). Thus, a collection of local sorghum varieties was carried out in 2009 in the four provinces of the region (Zondoma, Passoré, Yatenga and Lorum) and seed were stored in the gene bank of INERA Saria Research Station in Burkina Faso. Owing to the local differences between the villages and the interactions between genotypes and environments, the local sorghum varieties cultivated by farmers in the North region could show significant and useful biological traits.

This study was undertaken to assess the agromorphological diversity of local sorghum varieties from 13 villages of the North region of Burkina Faso to determine the structuration of the diversity and identify gene pools that could be of immediate use for the sorghum varietal improvement.

Collection site

The plant material of this study was collected from 13 villages of the four provinces in the North region of Burkina Faso (Zondoma, Passoré, Yatenga and Lorum) according to the method described by Kondombo et al. (2016). The villages are situated between latitudes 12°38' and 14°18' north and between 1°55 and 2°95 longitudes west (Figure  1). The climate of the region is a Sudano-Sahelian type, with a rainfall of 500 to 700 mm. The prevailing types of soil are Ferric-lixisol (39.4%) and lithosols on hardpan with gravelus (33.3%). The average dates of the beginning and the end of rainy season in the region are around 28^th June and 21^st September (National Direction of Meteorology, 2011).

Plant material

The characterization of agromorphological diversity was carried out with 150 local sorghum varieties from 2017 to 2018 at INERA Saria Research Station, situated in the North-Sudanian zone, at 12°16' North latitude and 2°09' West longitude, at 300 m altitude. The list of sampled villages is reported in the Table 1.

Experimental design and crop management

The experimental design was an alpha lattice with three repetitions. Each repetition includes 15 blocks of 10 varieties. The sowing was carried out on 12^th July, 2017. By repetition, each variety was sown on one row of 6 m length with spaces of 80 cm between rows and 20 cm between seed holes. The thinning was carried out 14 days after sowing, leaving one plant per seed hole. Mineral fertilizers NPK (14N-23P-14K-6S-1B) were applied at 100 kg/ha and urea at 50 kg/ha (46% of nitrogen) respectively 14 and 40 days after sowing. The utile rainfall for the trial was 444.2 mm.

Assessed traits

Thirteen qualitative traits and eleven quantitative traits were used to describe sampled varieties. (Tables 2a and 2b, respectively). Apart from the number of leaves recorded once a week from the thinning up to the appearance of the last leaf, the other traits were observed or measured at the stage of heading, milky grain stage, physiological maturity and after harvests (Table 2).

Regarding the quantitative traits, the variances analysis (ANOVA) were carried out at two levels to evaluate variety effect and the village effect on traits expression according to the following models: Y_ijk = m + t_i + g_j + r_jk + e_ijk for the variety and Y_ij = m + a_i + g_j + e_ij. for the village; m is general mean of the measured trait, t_i mean of the variety i, g_j effect of the repetition j, r_jk effect of block k in the repetition j, e_ijk residual effect for variety; a_i variety effect within the village and e_ij residual effect for village. The broad sense (H²) and the narrow sense (h²) heritabilities were calculated for each trait.

The structuration of agromorphological diversity was established by a Hierarchical Cluster Analysis (HCA) on the basis of the Euclidean distance according to Ward’s aggregation criterion. The groups from the HCA were characterized by a Factorial Discriminant Analysis (FDA) in order to identify the most discriminating traits of the groups. The FDA was used to check if the varieties were different from one village to another.

The ANOVA was performed using GenStat software version 14.2, whereas the MCA, HCA and FDA were carried out with the XLSTAT software version 2018.2 (Addinsoft, 2018).

Evaluation of qualitative traits

Almost all the 150 varieties assessed had leaf anthocyanin pigmentation (99.3%). The varieties with white leaf midrib (94.7%) and black glumes (77.3%) were the most common. The most differentiated modalities were observed on the pericarp colour (six modalities): the white pericarp was predominant (62.7%), followed by red pericarp (18.7%), orange (7.3%), brown (4.0%), grey (2.7%) and yellow (1.3%); while; 3.3% were mixtures. Seventy-eight percent of the varieties had grains without testa and 64.7% hadve presented good corneous to medium endosperm texture.

The modalities of the seven qualitative traits of the panicle have allowed identification for four main botanical races of sorghum and two intermediate races. The main races were: the guinea which represent 86.0% [two sub races: the gambicum (96.1%) and the margaritiferum (3.9%)], the caudatum (6.7%), the bicolor (0.7%) and the durra (0.7%). The intermediate races were: the guinea-caudatum (2.0%) and the durra-bicolor (4.0%). The structuration of the modalities of panicle traits is shown in the Figure 2, whereas the botanical races identified in the study is presented in the Figure 3. 

The results of this study are similar to those reported by Barro-Kondombo et al. (2010) who found 98.4% of anthocyanin sorghum varieties. Multi-scale investigations could provide more information on the nature of phenolic compounds  of  the  local  sorghum  varieties  in  Burkina Faso and the reasons why farmers prefer this trait. The results also showed a great presence of white pericarp varieties. Generally in Burkina Faso, white pericarp varieties are used in cooking to prepare thick porridge, a local dish widely consumed in rural families. On the other hand, red pericarp varieties are used in the preparation of local beer sold or served during traditional ceremonies. Some red pericarp varieties have a sweet grain and are sown to be consumed fresh during the lain period, as the earliest variety named Kapambga "sorghum whose grain is extracted fresh from the panicle".

Diversity and structuration of measured traits

The inter-varietal effect was very highly significant for all the eleven assessed traits, showing that the varietal diversity cultivated by the farmers was high. The greatest variations was observed with the cycle duration from sowing to heading and the average weight of 100 grains; these two traits strongly contribute to the differentiation of the local sorghum varieties assessed. At the village level in addition to the cycle duration and the weight of 100 grains, the number of leaves and the length of leaves were the traits which contributed to the varietal differentiation. The variability of the cycle within varieties was an expected result. In fact, the North region of Burkina Faso is an area more exposed to rainfall fluctuations. The beginning and the end of the rainy season is always uncertain, for that reason farmers integrate possible alternatives into their production strategies to maximize the chances of harvest. In case of an early beginning of rains the later-cycle varieties are the first to be sown and in the otherwise, short-cycle varieties are sown. Only farmers having farming areas around lowlands manage to sow late maturing sorghum varieties. Among the factors influencing crop diversity, climate effects are the most important Mercer and Perales (2010). Vom Brocke et al. (2010) also reported that the adaptability of the cycle was one of the criteria in sorghum growing areas in Burkina Faso.

Assessment of the agromorphological diversity of the 150 local sorghum varieties from the North region of Burkina Faso showed that the varietal diversity cultivated by farmers was high and largely dominated by sorghums of the guinea botanical race, which indicates that racial configuration of sorghums remains unchanged in Burkina Faso. In terms of management, the three groups from the (HCA) are different entities based on phenology. However, for the North region, the Group 3 which contains late maturing varieties (76.6 days NDH) characterized by low productivity would be more exposed to the risks of abandonments, due to the rainfall constraints and probably parasitic attacks like midge (Stenodiplosis sorghicola). For this plant material, molecular analysis would be useful to confirm gene pools. Moreover, it is important to confirm the agronomic performance of the early maturing varieties, their tolerance to sorghum pests and other pathogens, their grain quality and even their biochemical composition with the aim of their use in the varietal improvement program.

The authors have not declared any conflict of interests.

The authors are grateful to the Collaborative Crop Research Program of the McKnight Foundation which funded this study, as well as Dr Jacques Chantereau and Dr Batiseba Tembo Banda for reviewing the English of the final draft.