Research efforts in Burkina Faso are focused on providing suitable cotton varieties to mitigate the challenges of climatic variations, labor scarcity, and higher fiber quality required by the textile industry. The aim of this study is to investigate the potential, via prospection, collection, exchange, and introduction, of new germplasm material to generate genetic diversity and develop enhanced cultivars with needed characteristics for Burkina Faso’s future industry direction.
4.1. Potentialities of Crosses as Revealed by ANOVA, GCA, Heritability Estimates, and Correlations
The ANOVA revealed highly significant differences for all the agronomic traits studied, indicating considerable phenotypic diversity among the parents. Elite varieties from Burkina Faso (FK, BK1-BK2) expressed, in general, an average performance for almost all the agronomic traits evaluated. The other African materials exhibited useful potentialities, justifying their choice in this diallel to generate populations with a broad genetic base [
8]. Elite materials as well as other African types also showed some undesirable traits, such as late maturity. African cotton breeding programs used to select late flowering and maturing varieties, with a tendency to more vegetative growth [
2]. Among American breeding lines, TX5 showed the potential to reduce days to 50% flowering, decrease plant height, and increase seed size. TX1, TX2, and TX4 had higher seed indexes and TX3 was a good source for BW and FP compared to African varieties, which was unexpected. In the past, varieties introduced from the United States were found to have inferior FP compared to African varieties, which are among the best worldwide for fiber percent [
25,
26].
Combining ability describes the breeding value of parental lines and subsequently helps to identify desirable parents for hybridization and population development to expand the genetic variability for the selection of superior genotypes essential in a systematic breeding program [
11]. Therefore, GCA was analyzed, even though the present study was limited in the number of lines that could be included in the crossing scheme and showed that selected parent lines varied significantly (
Table 3). Results indicated that African lines FK37 (BK1), FK64 (BK2), and E53 (BK6), along with American lines TX6 and TX4, have interesting potential for improving components of maturity (Pfm, Opcm, and NFFB). Populations created from these groups of parents could potentially be used to develop early flowering and early boll maturing cultivars eventually. Conversely, E9 (BK4) and TX 307 (BK2) would contribute to developing late maturing cultivars. A previous study reported that American genotypes are good top-cross candidates to reduce the maturity period [
27]. The analysis of the ratio of mean squares, using the formula 2GCA/(2GCA + SCA) by Baker [
28], revealed that all the earliness traits are quantitatively inherited traits. Similar results were reported by Simon et al. [
29]. Other studies reported that days to 50% flowering is controlled by additive genes [
12,
30], or additive gene effects is most important for flowering [
13]. For traits related to the plant architecture (NVB, NFB, and PH), results indicate that most of the genotypes did not exhibit excessive vegetative growth except for E9 (BK4), Stam 59A (BK3), and TX 307 (TX2). American parents, except for TX1, were shorter in height compared to African parents, which were tall and vegetative. Within a context of climate changes, with water scarcity and cropping season shortening, most of the American lines studied here have characteristics to be exploited in breeding future Burkina Faso cultivars. Potential improvement of architectural traits, i.e., shorter plant height, early maturing, and lodging resistance, would allow mechanization and reduce the water demand. The ratio of mean squares indicated that these traits are all under the predominance of additive gene effects. Simon et al. [
29] and Raza et al. [
31] also mentioned that plant height is predominantly controlled by additive genes, while Vasconcelos et al. [
13] concluded on the prevalence of dominance effects for plant height in water stress condition.
Traits such as BVB, BFB, BP, and BW could be clustered as yield components. Most Burkina Faso materials exhibited positive GCA effects, while most American materials presented undesirable and negative GCA effects when evaluated in Burkina Faso. Only TX5 was similar, even better than African materials. It is obvious that, in their own conditions of production, most adapted cultivars are favored compared to foreign materials [
5]. However, this study suggests that crossing Burkina Faso materials with TX5 will likely result in hybrids with a greater number of bolls per plant, larger boll size, and, therefore, more yield. The analysis of the ratio of mean squares showed that BW is under additive effects, such as stated in the findings of Raza et al. [
31]. For bolls per vegetative branch, bolls per plant, and to some lesser extent bolls per fruiting branch, we concluded, in agreement with previous studies, on the presence of both additive and dominance gene effects, with additive being higher than dominance [
13,
23,
30]. For these traits, the selection should be based upon evaluation in several environments [
32].
Most Burkina Faso materials are better general combiners compared to American ones for fiber percent. Older accessions from the USDA NCGC (TX 307-TX2 and TX 294–TX1) reduced FP more than 2% in hybrids. However, newer breeding materials TX3 and TX4 are among the best contributors to improve FP. These two lines crossed with African materials (except for E9-BK4) would likely produce populations from which to select new cultivars with enhanced FP. Fiber percent is an effective breeding target in cotton, as it is solidly reported to be controlled by additive gene effects [
10,
12,
13,
23,
31], in agreement with results of the present study.
Heritability estimates (h
2) were classified as high (>0.50), medium (0.30–0.50), and low (<0.30) according to Bhateria et al. [
33]. In our study, two maturity traits (Pfm and Opcm) expressed both medium narrow-sense heritability (0.23 and 0.33) and broad-sense heritability (0.29 and 0.34), while NFFB was both low narrow-sense heritability (0.15) and broad-sense heritability (0.21). The selection for earliness based on NFFB appeared less effective than selection for other traits such as FP and SI [
10], requiring evaluation in multiple environments [
32]. Except for NVB (both medium narrow-sense and broad-sense heritability), architectural traits were both medium narrow-sense heritability and high broad-sense heritability. This means NFB and PH should be easily inherited in progenies; Lançon [
2] found similar results. Yield components expressed medium (BVB) to low (BFB, BP, and BW) narrow-sense heritability and medium (BW) to high (BVB, BFB, and BP) broad-sense heritability. These results concur with those reported by Cheatham et al. [
32], indicating complexity in breeding for these traits. FP and SI expressed both very high narrow-sense heritability (0.51 and 0.81) and broad-sense heritability (0.91 and 0.97). Lu and Meyers [
10] reported similar results for FP, confirming the relative ease to change this trait through breeding.
In general, many significant and positive correlations were observed among maturity, architectural, and production components. Lu and Meyers [
10] reported a positive and significant correlation between number of bolls per plant and boll weight with seed index. They also reported the number of bolls per plant to be significantly and negatively correlated with boll weight and seed index, which is not in accordance with our findings. Moreover, BW and FP were most often negatively correlated with the other traits. Correlation results indicated that it is theoretically possible to breed for many agronomic traits related to earliness, plant architecture, and yield simultaneously. With the other traits, it is possible to cause some potential antagonist effects; therefore, in an improvement program, the best parents for BW, FP, and SI might be exploited separately in different cross combinations.
4.2. Highlighting the Most Promising Combination of Crosses Based on SCA
SCA effects are usually used to identify the best cross combinations for hybrid production. SCA refers to dominance effects and non-additive interactions of genes, resulting from gene complementation between parents, as dominant or epistatic [
11].
When analyzing hybrids derived from parents with significant SCA, E9 (BK4) was the least favorable parent. Crosses between E9 and American parents (with E9 as the male parent), as well as between E9 and other Burkina Faso materials (with E9 as both the female and male parent), did not produce hybrids with desired characteristics. E9 is a primitive accession, known to be not as well adapted as the improved cultivars [
7,
8]. E9 appears to carry both negative trait associations into its hybrids and/or inhibit positive effects from the other parent. Cheatham et al. [
32] reported similar effects from an Australian converted wild accession, which tended to decrease most of the positive agronomic properties when it was used as a parent in crosses with American cultivars. The cross FK37 x E9, however, showed interesting effects by potentially reducing Pfm over seven days, increasing SI by 0.49 g and not influencing significantly the other parental means; it could be kept for an advanced improvement process by back crossing these desired genes into the FK37 parent.
Combinations within the remaining Burkina Faso materials exclusively, as well as within American materials exclusively, did not produce many hybrids with desired characteristics. When the elite FK37 cultivar was included as a male parent, SCA effects were significant only for SI and FP but in an antagonistic way (
Table 4). The elite FK64 cultivar resulted in significant SCA effects in hybrids only with two other parents, STAM 59A and E53, by decreasing important characteristics, i.e., NFB and BFB (−2.73 and −3.06, respectively) or SI and FP (−0.48 g and −0.77%, respectively). Parents affecting important characteristics in progenies were also reported by Simon et al. [
29]. Crosses between STAM 59A, E32, and E53 resulted in the only positive significant SCA effect, primarily for PH (
Table 4). The cross between the old cultivar STAM 59A and accession E32 appeared interesting, increasing PH (+23.36 cm) then NFB (+3.94) and BP (+6.34), while the other traits remained the same as the parents.
Crosses within American materials resulted in both positive and negative SCA effects, but no hybrid had enough interesting genetic gains to be considered promising for Burkina Faso. Previous studies reported that a narrow genetic base, with parents coming from a common origin or sharing common parents in their pedigree leading to insignificant or unfavorable SCA effects [
29,
32]. In other words, when selecting the genetic base to be used in diallel crosses, it is important to increase the frequency of favorable alleles in the traits of interest, allowing the development of top lines and genetic gain [
13]. Materials from Burkina Faso clustered in the same group in a study using molecular SSR markers [
7].
In the present study, the most promising hybrids predicted by SCA involved those resulting from crosses with Burkina Faso materials as male parents and American ones as female parents. The SCA effects revealed that seven hybrids, combinations between three Burkina Faso materials (FK37, FK64, and E32) and three American ones (improved breeding lines 15-10-610-7, 15-3-416, and 16-2-216FQ) showed potential for interesting improvements. Those parents in this study have higher parental values for the desired traits. FK37 × 15-10-610-7 F1 showed improved Pfm (−4.27 days) and FP (+0.87%), while PH was undesirably increased, and the remaining traits were not impacted. These two parents were also contributors in two other promising hybrids, FK37 × 15-3-416 and FK64 × 15-10-610-7. Hybrid FK37 × 15-3-416 is considered promising since no impact on most investigated traits was noticed, while important traits were improved, i.e., BW (+0.83 g). The low decline in FP (−0.70%) could be improved in backcrosses with FK37, which is good for this trait. Moreover, 15-3-416 is a desired parent in terms of earliness components, SI, and short stature.
FK64 × 15-10-610-7 improved the main traits desired of Burkina Faso’s future cultivars, i.e., reducing Opcm (−3.95 days) and increasing FP (+0.94%), while not significantly impacting other traits. This combination is interesting since FP and Opcm overall were found to be negatively correlated in this study. Cultivar FK64 is a parent in two other promising hybrids (FK64 × 16-2-216FQ and FK64 × 15-3-416). FK64 × 16-2-216FQ increased in FP (+1.75%), and it could be ranked among the best combination for FP; 16-2-216FQ was found to be a good parent to improve FP. However, PH was increased, and SI decreased, implying that backcrosses are required to address these issues. Abro et al. [
11], Khan [
12], and Ekenci and Basbag [
23] also recommended this process to improve hybrids after diallel crosses.
Hybrid FK64 × 15-3-416 increased SI (+0.62 g) and FP (1.19%) over the parent means while the other traits were maintained at parents’ average levels. This increase in SI will put the hybrid among the best for the trait, as 15-3-416 (10.38 g) was the highest ranked parent for SI and FK64 was also at a good level (9.08 g). This is an interesting observation since FP and SI can be negatively correlated.
E32 (BK5) was a parent in two of the most promising hybrids. E32 × 16-2-216FQ (TX3) increased BVB, PH, and FP by 2.01, 11.28 cm, and 2.06%, respectively, while maintaining the other traits the same as the parents. This hybrid is potentially the best for FP as its parents have higher parental values for this trait. E32 × 15-3-416 exhibited the widest range of desired SCA effects. Although PH undesirably increased (54.23 cm), the combination reduced Pfm over four days and increased BVB (+3.05 bolls), NFB (6.38 branches), BP (13.49 bolls), BW (1.53 g), SI (0.40 g), and FP (1.18%).
As a summary, the present study revealed that all hybrid combinations expressing desired SCA effects involved at least one or both high general combiners as parents, in accordance with many previous studies [
10,
11,
23,
29]. The possibilities to realize value from these populations exist through careful selection and carefully considering differences in germplasm materials and environmental effects [
10,
13,
29,
34,
35].