Diallel Crosses of Cotton (Gossypium hirsutum L.)—Enhancement of Fiber Properties in Future Cultivars for Burkina Faso

Abstract

The market value of cotton in exporting countries, such as Burkina Faso, depends on the quality of major fiber properties. A lack of variation among the genetic resources available in Burkina Faso hinders breeding progress to meet fiber quality demands in future cultivars. F₁ populations from a half diallel crossing scheme between lines developed at Texas A&M AgriLife cotton breeding program at Lubbock and germplasm accessed from Burkina Faso were evaluated for fiber property enhancement. Crosses exclusively within common origin materials did not result in highly enhanced hybrids. Hybrids within American materials expressed significant SCA effects undesirable for future cultivars in Burkina Faso. Five hybrids within the Burkina Faso material expressed significant SCA effects: two of them implicating FK37 as the better parent in transmitting superior targeted fiber traits to its progenies. Hybrid FK37xE9 significantly enhanced UHML, Str, UI and SFI but not Mic and Rd. Inter-program hybrids with at least one significant SCA effect were crosses with female American lines and male Burkina Faso parents. Hybrids E53x16-2-216FQ, E53x15-10-610-7 and E32x15-10-610-7 showed significantly improved Str and UI for the most economically important traits, which also include Mic, UHML, SFI, Rd and +b. Together, the GCA and SCA effects, heritability and correlations showed more additive than non-additive gene actions. Therefore, knowledge of the best combiners and hybrids could be used in the cultivar development process to enhance value with improved fiber characteristics.

1. Introduction

In Africa, the market value of cotton lies in the lint obtained after ginning by cotton companies that is sold for export, usually without any further processing. The cotton cultivated in Africa is mostly exported, and fiber quality is important [1]. On one hand, African cotton is competing globally with cotton growing regions where fiber quality is improving. On other hand, due to competition with synthetics (from man-made fibers), especially polyester, the world fiber market requires cotton to meet quality standards within traditional parameters (fiber length, grade, color and micronaire index), as well as characteristics such as cleanliness, homogeneity, strength, uniformity, maturity, fineness, percentage of short fibers, elasticity, dyeability, etc., such that cotton fiber can meet the spinning process requirements [1].

Through a short-term response in terms of cleanliness, programs and sensitization initiatives were implemented to solve the problem of contamination by non-plant materials. From a medium- to long-term perspective, it is recognized that breeding progress in Africa has declined in recent years as cultivars otherwise considered obsolete are not replaced with improved genetic gain from selection. In current and future cultivar development, cotton breeders must systematically evaluate ten or more characteristics whose fiber values are expected at specific levels or intervals to optimize the economic benefits at the point of sale.

The future cultivar development in Burkina Faso is nationally concentrated on objectives to meet the evolving cotton cultivation methods and fiber quality requirements of the textile industry. These objectives include early crop maturity and water use efficiency while maintaining and continuing to improve the high lint turnout and excellent fiber quality reputation of African cotton. Two initiatives to consider in breeding objectives are the potential future mechanization and the organic production methods adopted, mostly by women, in Burkina Faso.

Cultivar development rests on whether appropriate genetic variability exists, and the methods and tools used for crop improvement. If the end goal, in classic as well as in specific crop improvements, is to create novel allele combinations via varying crossing schemes, a major challenge is the strategy for the selection of parents that will produce the expected result in desired cultivars. The general approach is to cross genetically unrelated cultivars, but information on how to select these parents in cotton is critical [2].

Analysis of a diallel crossing scheme can reveal what type of gene action is influencing the phenotype of various plant characteristics. The diallel crossing scheme is designed to help understand the inheritance of traits to facilitate the development of genetic populations [3]. Studies using diallel mating schemes have been reported for multiple crops, including cotton [4,5]. The half-diallel cross creates all possible combinations among parents without reciprocal crosses [6,7]. Half-diallel and the full-diallel (direct and reciprocal) crosses have been used in cotton to analyze the general combining ability (GCA) and specific combining ability (SCA) of germplasm for improving agronomic and fiber properties [8,9,10]. The GCA indicates the overall parental genetic influence on progeny traits due to additive effects. The SCA refers to the non-additive gene (dominant or epistatic) effects of a specific hybrid [5,11,12]. Therefore, a high SCA and at least one parent with a high GCA can identify hybrids favorable for the studied breeding objectives [5]. The diallel analysis can also assist the cotton breeder in developing a proper strategy to evaluate breeding material likely to provide meaningful germplasm improvement [9,12,13,14].

This study was developed to evaluate the potential for cotton lines developed at Texas A&M AgriLife Research in Lubbock (LREC) to address the objectives needed in future cultivars for Burkina Faso [8]. The mating design used two groups of parents, six lines from Africa and six lines from the United States, chosen for how their fiber properties might meet Burkina Faso’s quality objectives. Previously, we used the same populations to analyze the improvement of agronomic traits [8]. Therefore, now we aim (i) to highlight the breeding potential of the developed population for fiber property traits and (ii) to identify early in the breeding process the most promising combination of crosses to enhance the major marketing fiber properties (micronaire, length, strength, and color). The hypothesis is that hybrids with parents developed in diverse growing regions will introduce positive genetic variation for breeding fibers with improved major properties in future cultivars.

2. Materials and Methods

2.1. Plant Materials

The cotton lines used as parents included two cultivars from Burkina Faso, FK37 (BK1) and FK64 (BK2), and one originating from Togo, STAM 59A (BK3) (

Table 1. Designation and origin of the 12 cotton lines used in a half-diallel crossing scheme.

Line Designation	Code	Origin	Type
FK37	BK1	Burkina Faso	Improved Cultivar
FK64	BK2	Burkina Faso	Improved Cultivar
STAM 59A	BK3	Togo Introduction	Improved Cultivar
E9	BK4	Burkina Faso	Germplasm Collection Accession
E32	BK5	Burkina Faso	Germplasm Collection Accession
E53	BK6	Burkina Faso	Germplasm Collection Accession
TX 294	TX1	United States	Germplasm Collection Accession
TX 307	TX2	United States	Germplasm Collection Accession
16-2-216FQ	TX3	United States	LREC Breeding Line
15-10-610-7	TX4	United States	LREC Breeding Line
16-2-418BB	TX5	United States	LREC Breeding Line
15-3-416	TX6	United States	LREC Breeding Line

). FK37 and FK64 are widely cultivated and well adapted to Burkina Faso’s local production conditions [8]. Three lines, E9 (BK4), E32 (BK5) and E53 (BK6), accessed from the local germplasm collection of Burkina Faso were selected based on their good agromorphological or fiber properties [15,16]. Two lines, TX 294, PI 165244 (TX1) and TX 307, PI 165390 (TX2), were from the USDA National Cotton Germplasm Collection (NCGC) and four, 16-2-216FQ (TX3), 15-10-610-7 (TX4), 16-2-418BB (TX5) and 15-3-416 (TX6), were unreleased breeding lines from Texas A&M AgriLife Research in Lubbock (LREC) that were selected for their characteristics, which have the potential to meet the requirements of future cultivars for Burkina Faso [8]. The line designation and codes used in the results, along with germplasm origin and type, are shown in Table 1.

2.2. Crosses

Hybrid seeds were generated at the LREC greenhouse facility using a half-diallel crossing design [8]. Seeds from the twelve lines were treated as in [8], with an 80 °C hot water bath (Precision 180 series, SN. 698091449; Jouan, Incorporation, Winchester, VA, USA) for 90 s to enhance germination [17] and planted in 0.5 L cups filled with potting soil media. Ten days after planting, the seedlings were transplanted to 19 L pots with potting soil (Berger Peat Moss Ltd., Saint-Modeste, QC, Canada) mixed with a slow release 15-9-12 fertilizer (Fertilizer Everiss NA Inc., Dublin, OH, USA). Seedlings were tested using ELISA strips AS-069, AS-175, AS-096 (Envirologix Inc., Portland, ME, USA) to detect and mitigate the unintentional presence of deregulated genetic engineered traits. Eight plants each of TX1–TX6, eight plants each of BK1 and BK4, eight plants of BK3, ten plants of BK2, five plants of BK5 and two plants of BK6 survived for crossing. Flower buds were hand-emasculated one day prior to anthesis and pollinated the next day according to the half-diallel crossing scheme. Methods to prevent alien pollen contamination included gluing the candle petals together prior to emasculation and protecting the stamen with a paper straw (Artstraws Ltd., Swansea, Wales, UK) before and after pollination. Small amounts of F₁ seeds of 66 hybrids were harvested [8].

2.3. Field Experiment

The parents and hybrids were planted in a randomized complete block design, with three replications, in four-row 80 cm plots and 40 cm plant spacing at the Agricultural Research Station of Farako-Bâ in June 2019. This site is in the western cotton production area of Burkina Faso between the isohyets 800 and 1200 mm at 405 m altitude, 4°200′ O longitude and 11°060′ N. The plots were hand-planted and thinned to two plants hill⁻¹. Fertilization and pest and weed control were conducted according to the Burkina Faso national recommendations for conventional cotton production [8]. Ten plants from the two middle rows of each plot were randomly selected to collect data on fiber characteristics, as suggested by Abro et al. [4].

2.4. Laboratory Testing and Measurement of Fiber Properties

Seed cotton collected from each plot was hand-harvested and ginned with a laboratory roller gin. Then, 100 to 150 g of fiber was sub-sampled from each plot. The current ASTM D5867 (Standard Test for Measurement of Physical Properties of Cotton Fiber by High Volume Instruments) [18] method was used to measure the fibers’ technological characteristics on a SOFITEX (Société Burkinabè des Fibres et Textiles) integrated measurement chain (ICM). The ambient conditions in the laboratory for sample conditioning and testing met the requirements of the current ASTM standard (21 °C ± 1 °C temperature and 65% ± 2% relative humidity). The reference cottons or standard cottons used for calibration were: Universal Standard Short/Weak Calibration Cotton (ex: Universal Standard Short Weak 36660, with expiration date July 2024), Universal Standard Long/Strong Calibration Cotton (ex: Universal Standard Long/Strong Calibration Cotton 36587, with expiration date June 2024), Universal Standard Micronaire Calibration Cotton G_m-37262, low Mic (2.68) and Universal Standard Micronaire Calibration Cotton A_u-33612, high Mic (5.52).

The fiber properties investigated were micronaire (Mic), length in mm (UHML), uniformity index as a percent (UI), short fiber index % below 12.7 mm (SFI), strength in g tex⁻¹ (Str), elongation percent (Elo), degree of reflectance (Rd) and Hunter’s chromaticity coordinate for yellowness (+b).

2.5. Data Analysis

Analysis of variance on the fiber properties was conducted using the SISVAR 5.1 Build 72 software (Federal University of Lavras, Lavras, MG, Brazil). The Scott–Knott test at a 0.05 threshold level separated means. A fixed model was used for data analysis. The data were further analyzed to determine specific combining ability effects according to Griffing’s [12] “Method II ’Model I”, using DIAL 98 software, revised 22 September 2002 (Prof. Ukai Yasuo, University of Tokyo, Tokyo, Japan). The same software correlation analysis was used for GCA effects according to Griffing’s [12] “Method II ‘Model II”.

The relative importance of the GCA and SCA effects on the inheritance of the studied traits was evaluated using the formula:

2GCA/(2GCA + SCA)

(1)

The closer this ratio is to 1, the more important the additive gene effects are in the inheritance of the trait [3,19].

DIAL 98 Software and Hayman’s [20] method was used to estimate the broad sense and narrow sense heritability with the following formula:

h²_n: (σ²_A/σ²_P) and h²_b: (σ²_G/σ²_P)

(2)

where h²_n is the narrow sense heritability, h²_b is the broad sense heritability, σ²_A is the additive variance, σ²_G is the genotypic variance and σ²_P is the phenotypic variance.

3. Results

3.1. ANOVA of Fiber Property Traits among Parents and Mean Squares

The fiber property means (

Table 2. Fiber quality trait means of 12 cotton lines, 6 originating from Burkina Faso and 6 from the United States.

Line	Mic	UHML	UI	SFI	Str	Elo	Rd	+b
FK37 (BK1)	3.80 a2	27.58 a2	80.39 a1	13.03 a2	28.34 a2	5.91 a1	78.32 a3	7.59
FK64 (BK2)	3.85 a2	30.24 a2	81.33 a2	10.38 a1	30.01 a2	6.21 a1	78.49 a3	8.06
STAM 59A (BK3)	4.06 a2	27.60 a2	82.14 a2	10.59 a1	29.70 a2	6.49 a1	78.19 a3	7.81
E32 (BK5)	3.90 a2	27.60 a2	83.05 a2	9.95 a1	29.75 a2	5.94 a1	76.44 a2	8.75
E53 (BK6)	3.86 a2	28.71 a2	82.40 a2	10.29 a1	29.49 a2	6.49 a1	75.67 a2	8.00
E9 (BK4)	4.04 a2	26.45 a2	80.17 a1	13.71 a2	25.59 a1	6.35 a1	71.80 a1	6.22
TX 294 (TX1)	2.93 a1	26.80 a2	78.22 a1	16.18 a2	24.64 a1	5.75 a1	78.48 a3	7.28
TX 307 (TX2)	4.96 a3	22.18 a1	78.07 a1	16.61 a2	24.24 a1	8.42 a3	75.31 a2	8.16
15-10-610-7 (TX4)	2.96 a1	28.73 a2	78.86 a1	14.88 a2	29.12 a2	6.96 a2	76.67 a2	7.39
16-2-216FQ (TX3)	3.78 a2	27.69 a2	82.64 a2	11.19 a1	29.66 a2	7.23 a2	77.37 a3	8.02
15-3-416 (TX6)	3.24 a1	29.59 a2	81.35 a2	11.00 a1	30.15 a2	6.54 a1	77.39 a3	7.98
16-2-418BB (TX5)	3.86 a2	26.66 a2	81.47 a2	12.21 a1	26.07 a1	7.28 a2	79.19 a3	7.62
Mean	3.77	27.49	80.84	12.50	28.06	6.63	76.94	7.74
Cv (%)	9.97	5.89	1.57	16.90	7.42	7.64	1.66	11.66
p-value	0.000 **	0.001 **	0.000 **	0.003 **	0.005 **	0.000 **	0.000 **	0.235

Mic = Micronaire; UHML = Fiber length (mm); UI = Uniformity (%); SFI = Short fiber index (%); Str = Strength (g tex⁻¹); Elo = Elongation (%); Rd = Reflectance; +b = Yellowness index. Means within a column followed by the same letter (a) and index number (a₁ or a₂ or a₃, etc.) are not significantly different according to the Scott–Knott test. Values followed by ** are significantly different at p ≤ 0.01, respectively.

) and analysis of variance (

Table 3. Half-diallel general and specific combining ability analysis of the fiber quality traits of 12 cotton lines, 6 originating from Burkina Faso and 6 from the United States.

	Df	Mic	UHML	UI	SFI	Str	Elo	Rd	+b
GCA	11	4.54 **	54.99 **	16.21 **	30.04 **	68.08 **	4.14 **	23.27 **	3.71 **
SCA	66	0.16	1.78	2.55 **	4.34 *	6.02 **	0.28 **	2.35	0.53 **
Error (Residuals)	154	0.13	1.44	1.49	3.01	3.25	0.14	1.85	0.29

Mic = Micronaire; UHML = Fiber length (mm); UI = Uniformity (%); SFI = Short fiber index (%); Str = Strength (g tex⁻¹); Elo = Elongation (%); Rd = Reflectance; +b = Yellowness index. Values followed by * or ** are significantly different at p ≤ 0.05 or p ≤ 0.01, respectively.

) showed that significant differences (at least p < 0.05) were observed among the parents for all the investigated traits except +b.

The trait means (Table 2) show the Mic values from all the Burkina Faso lines were within the premium range, i.e., between 3.8 and 4.2. Among the US samples, TX3 (3.78) and TX5 (3.86) fall within the premium range, whereas the others are below (TX1, TX4, TX6) or above (TX2) it. For the UHML, all the material studied except TX2 (22.18 mm) exhibited similar values, ranging from 26.45 mm (BK4) to 30.24 mm (BK2). For Str, all material from Burkina Faso except BK4 (25.59) and those from US except TX1 (24.64 g tex⁻¹), TX2 (24.24 g tex⁻¹) and TX5 (26.07 g tex⁻¹) showed the desired good values (28.34 to 30.15 g tex⁻¹). TX1 (5.75%) and TX6 (6.54%) joined all Burkina Faso parents (5.91 to 6.49%) in the lower elongation group with other US parents (6.96 to 7.28%) being better and TX2 having significantly higher elongation (8.42%) than all the parent lines. The reflectance (Rd) color grade trait divided the parents into three groups: BK4 (71.80) had the least brightness; intermediate reflectance was found for two samples from Burkina Faso, BK5 (76.44) and BK6 (75.67), and the US lines TX2 (75.31) and TX4 (76.67); and the highest reflectance was found for the group that included the Burkina Faso cultivars BK1 (78.32), BK2 (78.49) and BK3 (78.19) plus four US entries, TX1 (78.48), TX3 (77.37), TX5 (79.19) and TX6 (77.39).

Analysis of variance for the half-diallel crossing (Table 3) shows that the GCA effects were significant (p ≤ 0.01) for all the fiber property variables studied, indicating that additive gene action was prevailing over dominant or epistatic expression for these traits. SCA was found to be significant for five out of the eight variables studied, including one, the SFI, at (p ≤ 0.05) and four at a 1% threshold, including the UI, Str, Elo and +b, revealing an important role for non-additive gene effects.

3.2. Analysis of Combining Ability

3.2.1. GCA of Parents

Only six parents imported to Texas with a permit from Burkina Faso and six parents from Texas for F₁ seed production in Texas and evaluation in the target environment of Burkina Faso were available for this project. Moreover, the environmental evaluation was limited due to seed volume [8]. More genotypes would have strengthened confidence in the results; however, GCA analysis did indicate even among these 12 parents a robust range and significant effects, both positive and negative, for the fiber traits evaluated (

Table 4. General combining ability (GCA) of the fiber quality traits of 12 cotton lines originating from Burkina Faso and the United States.

Parent	Mic	UHML	UI	SFI	Str	Elo	Rd	+b
FK 37 (BK1)	−0.10	0.61 *	−0.22	−0.03	0.64 *	−0.34 **	0.57 *	0.11
FK 64 (BK2)	0.09	1.01 **	−0.06	−0.51	0.45	−0.17 *	0.27	0.23 *
STAM 59A(BK3)	0.16 *	0.32	0.36	−0.56	0.88 *	−0.14 *	0.39	0.07
E32 (BK5)	−0.06	−0.14	0.95 **	−0.98 **	1.12 **	−0.31 **	−0.03	0.38 **
E53 (BK6)	0.09	0.49 *	0.39	−0.66 *	0.34	−0.08	−0.33	0.11
E9 (BK4)	0.27 **	−0.59 *	0.17	0.05	−0.77 *	−0.02	−1.78 **	−0.69 **
TX 307 (TX2)	0.71 **	−3.19 **	−0.54 *	0.77 *	−2.58 **	0.61 **	−1.01 **	−0.06
TX 294 (TX1)	−0.52 **	−0.50 *	−1.46 **	2.20 **	−1.96 **	−0.46 **	0.58 **	−0.43 **
15-10-610-7(TX4)	−0.47 **	1.21 **	−0.43 *	0.48	1.22 **	0.13 *	0.64 **	−0.10
16-2-216FQ (TX3)	−0.11	0.22	0.57 *	−0.42	1.07 **	0.21**	−0.23	0.23 *
15-3-416 (TX6)	−0.19 **	0.59 *	0.09	−0.13	0.57 *	0.22 **	0.39	0.09
16-2-418BB (TX5)	0.12	−0.02	0.18	−0.21	−1.00 **	0.34 **	0.55 *	0.07

Mic = Micronaire; UHML = Fiber length (mm); UI = Uniformity (%); SFI = Short fiber index (%); Str = Strength (g tex⁻¹); Elo = Elongation (%); Rd = Reflectance; +b = Yellowness index. Values followed by * or ** are significantly different at p ≤ 0.05 or p ≤ 0.01, respectively.

).

The favorable Mic range for spinning is a value between 3.8 and 4.2, and most parents are good for this variable. The US cotton market loan values apply a slight premium for Mic values between 3.8 and 4.2, with 3.5–3.7 and 4.3–4.9 neutral and below 3.5 and above 4.9 discounted. The GCA analysis shows that the American parents TX1 (−0.52), TX4 (−0.47) and TX6 (−0.19) can significantly lower the Mic trait, whereas BK3 (0.16) could have a positive influence in the progeny. The parents TX4 (1.21), BK2 (1.01) and to a lesser extent BK1 (0.61), TX6 (0.59) and BK6 (0.49) were the best parents for positively impacting the UHML in the progeny. On the other hand, BK4 (−0.59), TX1 (−0.50) and especially TX2 (−3.19) were the worst parents for improving fiber length in the progeny. For UI, BK5 and TX3 resulted in more uniform fibers in the progeny, whereas TX1 (−1.46), TX2 (−0.54) and TX4 (−0.43) had a potentially negative influence on UI improvement in the progeny. BK5 (−0.98) and BK6 (−0.66) improved the SFI in the progeny, as smaller percentage of short fibers is desirable. On the other hand, TX2 (0.77) and especially TX1 (2.20) increased the SFI in the progeny, which can significantly deteriorate the spinning performance.

Three BF parents (BK5, BK1 and BK3) and three US parents (TX4, TX3 and TX6) showed a positive impact on Str by increasing fiber tenacity in the progeny. BK4, TX5, TX1 and especially TX2 had a negative impact on the fiber tenacity of their progenies. The lower Elo in parents from Burkina Faso had a negative impact on Elo in their progeny, particularly BK3 (−0.14), BK1 (−0.34), BK2 (−0.17) and BK5 (−0.31). On the other hand, all US parents, except TX1 (−0.46), are expected to significantly improve fiber elasticity in new population development. Half of the parents could significantly influence reflectance in their offspring. BK1 (0.57), TX1 (0.58), TX4 (0.64) and TX5 (0.55) had a positive influence, whereas BK4 (−1.78) and TX2 (−1.01) had a negative influence. As far as colorimetry (yellowness) is concerned, the parents BK2 (0.23), BK5 (0.38) and TX3 (0.23) led to an increase (undesirable) of the value in progenies, whereas BK4 (−0.69) and TX1 (−0.43) could potentially improve the color grade of the progenies.

3.2.2. SCA of Hybrids

Of the 66 hybrids, 25 showed a significant SCA effect for at least one fiber quality trait (

Table 5. Specific combining ability (SCA) effects for fiber quality traits of 25 cotton hybrids with parental lines from Burkina Faso and the United States. (Only hybrids with at least one significant trait are included).

Hybrid	Mic	UHML	UI	SFI	Str	Elo	Rd	+b
FK37xTX 307 (BK1xTX2)	0.15	−0.02	0.26	−2.30 *	0.76	0.32	−1.07	0.06
FK37x15-3-416 (BK1xTX6)	−0.44 *	0.49	−1.19	0.83	1.75	−0.16	0.81	0.44
E32xTX 307 (BK5xTX2)	0.24	−1.18	0.82	−0.46	−1.36	−0.4	−1.53 *	0.15
E32xTX 294 (BK5xTX1)	−0.39	−0.08	−0.6	0.99	−0.32	0.14	−1.21	−0.95 **
E32x15-10-610-7 (BK5xTX4)	−0.2	0.59	1.47 *	−1.4	0.44	0.02	1.11	−0.22
E32x16-2-216FQ (BK5xTX3)	0.18	−0.25	0.59	−0.9	2.63 *	0.26	0.13	0
E53xTX 307 (BK6xTX2)	−0.11	0.3	−0.39	0.8	0.78	−0.57 *	0.99	−0.98 **
E53x15-10-610-7 (BK6xTX4)	−0.08	0.64	0.42	−0.3	2.65 *	0.15	0.1	−0.32
E53x16-2-216FQ (BK6xTX3)	−0.12	0.97	0.5	−0.23	2.13 *	−0.42 *	1.03	−0.42
E9xTX 307 (BK4xTX2)	−0.2	0.97	0.6	−0.56	1.75	−0.51 *	0.84	0.04
E9xTX 294 (BK4xTX1)	0.16	−0.5	−0.27	0.21	−1	−0.03	−0.11	−0.95 **
E9x15-10-610-7 (BK4xTX4)	−0.09	1.15	0.63	−1.93 *	1.28	−0.02	1.21	0.16
E9x15-3-416 (BK4xTX6)	0.55 *	−1.82 *	0.24	1.75	−2.05 *	1.35 **	−0.14	−1.08 **
E9x16-2-418BB (BK4xTX5)	0.31	0.82	1.05	−1.14	1.07	−0.73**	0.91	0.05
FK37xSTAM 59A (BK1xBK3)	0.15	0.16	1.13	−0.48	1.34	−0.02	0.66	−0.82 *
FK37xE9 (BK1xBK4)	−0.07	1.43 *	2.30 **	−2.44 *	2.48 *	−0.12	0.72	0.3
STAM 59AxE9 (BK3xBK4)	0.19	0.08	0.04	−0.42	0.64	0.16	−0.32	0.75 *
E32xE53 (BK5xBK6)	0.23	−0.68	−1.19	−0.04	−0.77	0.3	−0.42	0.65 *
E32xE9 (BK5xBK4)	−0.63 **	−0.4	−0.38	1.42	−0.74	0.13	2.31 **	−0.83 *
TX 307x15-10-610-7 (TX2xTX4)	0.1	−0.88	0.34	−0.14	−3.53 **	−0.07	0.6	−0.19
TX 307x16-2-216FQ (TX2xTX3)	−0.06	0.43	1.39 *	−2.16 *	1.29	−0.09	0.61	−0.16
TX 307x15-3-416 (TX2xTX6)	0.22	−0.69	1.79 *	−1.35	−1.81	−0.33	0.47	−0.48
TX 294x16-2-216FQ (TX1xTX3)	−0.32	−1.04	−1.04	1.3	−2.67 *	−0.48 *	−1.19	0.18
15-10-610-7x15-3-416 (TX4xTX6)	0.09	−0.6	−0.44	1.56	−0.92	−0.44 *	0.32	0.05
16-2-216FQx16-2-418BB (TX3xTX5)	0.16	0.31	−0.28	−0.05	−2.26 *	0.16	−0.97	−0.14

Mic = Micronaire; UHML = Fiber length (mm); UI = Uniformity (%); SFI = Short fiber index (%); Str = Strength (g tex⁻¹); Elo = Elongation (%); Rd = Reflectance; +b = Yellowness index. Values followed by * or ** are significantly different at p ≤ 0.05 or p ≤ 0.01, respectively.

). All variables showed significant SCA effects for at least two hybrids; the traits Str and +b showed more combinations with positive SCA effects (eight combinations each), whereas the UHML and Rd showed the least (two combinations each). Most crosses with a significant SCA for several traits (>2) involved the genotype E9 (BK4) as the male or female parent.

3.2.3. SCA Effects of Hybrids for the Major Fiber Characteristics (Mic, UHML, Str and +b)

The SCA effect for the Mic trait was significant for three hybrids. The Mic would be potentially higher in hybrid BK4xTX6 (0.55) and potentially lower in hybrids BK1xTX6 (−0.44) and BK5xBK4 (−0.66) compared with the average of their respective parents. The SCA effect for the UHML was positive for two hybrids, both involving BK4 as either the male or female parent. The effect on the UHML was positive for BK1xBK4 (1.43), indicating an improvement in fiber length compared with the average of the parents, whereas it was negative for the hybrid BK4xTX6 (−1.82), indicating a decline in this case. Fiber strength is one of the traits for which more hybrids have significant and high SCA values. Hybrids with positive SCA values (BK5xTX3, BK6xTX4, BK6xTX3, BK1xBK4) and those with negative SCA values (BK4xTX6, TX2xTX4, TX1xTX3 and TX3xTX5) always involved TX3 (male or female parent), BK4 (male or female parent) or TX4 (as female parent). The yellowness index (+b) in the progeny was influenced by many parents. Six hybrids (BK5xTX1 (−0.95), BK6xTX2 (−0.98), BK4xTX1 (−0.95), BK4xTX6 (−1.08), BK1xBK3 (−0.82), BK5xBK4 (−0.83)) were less yellow (better color grade) than the average of the parents, whereas two hybrids, BK3xBK4 (0.75) and BK5xBK6 (0.65), produced a +b value above the average of their parents, which is not desired.

3.2.4. SCA of Hybrids for Other Fiber Properties (UI, SFI, ELO and Rd)

Significant and positive SCA effects for UI were observed in the following four hybrids, BK5xTX4 (1.47), BK1xBK4 (2.30), TX2xTX3 (1.39) and TX2xTX6 (1.79); these hybrids were found to be the best combinations for fiber uniformity. All the hybrids with significant SCA effects for the SFI expressed negative effects, meaning that these hybrids performed under the mean average of the parents even when the parents appeared desirable for this trait. Seven hybrids exhibited significant SCA effects for Elo but only one, BK4xTX6 (1.35), was a positive and desired effect, whereas the remaining four were negative and less desired hybrids for fiber elasticity. The hybrids BK5xTX2 (−1.53) and BK5xBK4 (2.31) expressed significant SCA effects for Rd, the former expressing a negative and the latter expressing a positive desired effect.

3.3. Estimate of Heritability

The variances (additive, genotypic and phenotypic) and heritability (narrow and broad sense) are presented in

Table 6. Heritability estimates for fiber quality traits of 12 cotton lines originating from Burkina Faso and the United States.

Fiber Quality Traits	σ2A	σ2G	σ2P	h2n (σ2A/σ2P)	h2b (σ2G/σ2P)
Mic	0.08	0.08	0.21	0.37	0.37
UHML	1.35	1.35	2.79	0.48	0.48
UI	0.72	0.72	2.21	0.33	0.33
SFI	1.29	1.29	4.29	0.30	0.30
Str	0.47	0.69	3.94	0.12	0.17
Elo	0.18	0.19	0.34	0.53	0.57
Rd	1.08	1.08	2.93	0.37	0.37
+b	0.02	0.03	0.32	0.05	0.09

σ²_A = Additive variance; σ²_G = Genotypic variance; σ²_P = Phenotypic variance; h²_n = Heritability in narrow sense; h²_b = Broad-sense heritability. Mic = Micronaire, UHML = Fiber length (mm), UI = Uniformity (%), SFI = Short fiber index (%), Str = Strength (g tex⁻¹), Elo = Elongation (%), Rd = Reflectance, +b = Yellowness index.

. The additive variance for the UHML (1.35), SFI (1.29) and Rd (1.08) was above one but under two, whereas the rest of the traits were below one. A similar trend was observed for the genotypic variance, with almost the same values except for Str. The UHML, SFI and Rd were under two, whereas the Mic, UI, Str, Elo and +b was under one.

Because additive variance was close to genotypic variance, narrow-sense heritability was nearly equal to broad-sense heritability, except to a lesser extent for +b (0.05 versus 0.09), Str (0.12 versus 0.17) and Elo (0.53 versus 0.57). Two traits (Str and +b) expressed the lowest values, whereas Elo expressed higher heritability. For most traits investigated, the heritability was ≥30. This study found the GCA effects to be higher than the SCA effects, which indicates much additive gene action and little non-additive gene action.

3.4. Correlation Effects

Relationships among the GCA effects for the observed traits that were specific to the limited number of lines in this study are summarized in

Table 7. Correlation coefficients between GCA effects on the fiber quality traits of 12 cotton lines originating from Burkina Faso and the United States.

Fiber Quality Traits	Mic	UHML	UI	SFI	Str	Elo	Rd	+b
Mic	1
UHML	−0.66 *	1
UI	0.28	0.25	1
SFI	−0.29	−0.38	−0.95 **	1
Str	−0.41	0.80 **	0.62 *	−0.69 *	1
Elo	0.52	−0.46	0.09	−0.05	−0.29	1
Rd	−0.66 *	0.59 *	−0.20	0.08	0.36	−0.31	1
+b	0.01	0.30	0.56	−0.65 *	0.56	0.04	0.46	1

Mic = Micronaire, UHML = fiber length (mm), UI = Uniformity (%), SFI = Short fiber index (%), Str = Strength (g/tex), Elo = Elongation (%), Rd = Reflectance, +b = Yellowness index. Values followed by * or ** are significantly different at p ≤ 0.05 or p ≤ 0.01, respectively.

. The most significant correlations are the negative ones between UHML and Mic (−0.66), Rd and Mic (−0.66), SFI and UI (−0.95), Str and SFI (−0.69) and +b and SFI (−0.65), whereas there are three significant positive correlations, Str and UHML (0.80), Str and UI (0.62) and Rd and UHML (0.59). The most solid significant correlations were positive (0.80) or negative (−0.95), i.e., the correlations between Str and UHML and SFI and UI, respectively. The Elo trait did not correlate with any other trait in this study, whereas +b was only correlated with SFI (−0.65).

4. Discussion

Global research efforts in Burkina Faso focus on providing suitable cotton cultivars to mitigate challenges from climatic variations, labor scarcity and the higher fiber quality requirements by the textile industry. Burkina Faso has been looking for fiber quality enhancement since Bt cotton cultivation was suspended in 2016; Bt varieties underperformed compared with their conventional isogenics on major fiber traits, including UHML and Str [21]. The present study, using a half-diallel analysis involving six genotypes from Burkina Faso and six US-sourced genotypes, was an effort to produce cotton hybrids expressing potential genetic variations for enhanced fiber quality.

4.1. GCA of Parents

Significant differences among genotypes for all the fiber characters studied indicate the promise of considerable genetic variation in the experimental materials. The lines in this diallel study were limited in number because of international exchange logistics and were chosen for specific desirable characteristics as well as the potential for introducing positive variation. Other authors have previously reported similar results for genotypes in diallel studies [6,22,23,24] regarding initial parent performance, potential to influence progeny and explaining diversity among the lines.

The Burkina Faso elite cultivar FK64 (BK2) was a superior combiner for the main fiber marketing properties of UHML and Str, whereas the American line 15-3-416 (TX6) was closest to the fiber quality values for the improved UHML (29.59 mm) and Str (30.15 g tex⁻¹) objectives. Older accessions from the USDA NCGC, TX 307 (TX2) and TX 294 (TX1), as well as the genotypes E32 (BK5) and E9 (BK4) showed significant positive performance for properties of less economic interest but that are important technical attributes, i.e., UI and SFI. TX 307 (TX2) was superior for Mic (4.96) and Elo (8.42) traits and inferior for UHML, the SFI and Str. Most African cultivars recorded similar or superior average performances over the American parents used in this study for all the parameters except for Elo.

According to the Interprofessional Cotton Association of Burkina Faso (AICB) regarding preferred fiber marketing traits [25], the GCA and SCA results from this study indicate the American genotypes used would not be suitable for reaching the desired fiber quality in new cultivars for Burkina Faso. The AICB conference stipulated that for significant enhancement of the fiber properties in Africa new sources of genes need to be considered, including those not yet used in African breeding programs, e.g., Asiatic and some Australian genetic resources reported to possess genes that can improve fiber quality [26].

The GCA effects for the average performance of parents in hybrid combinations were significant for all eight fiber properties under study. The ratio 2GCA/(2GCA + SCA) [19] was near to 1, implying that the GCA was more important than the SCA in this study. For the Mic, UHML and Rd traits, the GCA variance alone was significant, revealing the important role of additive gene effects. For other properties, both the GCA and SCA variances were significant, indicating possibly both additive and non-additive gene effects.

Several previous studies on gene actions on fiber properties have focused on fiber strength, length, fineness (estimated by micronaire) and elongation because they are seen as the measurable properties that influence spinning performance [6,13,27,28]. Our results were consistent with those from Meredith and Bridge [29], Singh and Chahal [30], Karademir and Gencer [13], Bolek et al. [22] and Khan et al. [31] as they showed higher GCA effects than SCA effects, indicating large additive gene action against little non-additive gene action in the control of the UHML, UI, SFI, Elo and +b. Raiz et al. [32] reported both GCA and SCA significant effects, with the GCA values being somewhat superior to the SCA values. Palve [27] found only GCA variance significant but not SCA variances for the main fiber properties in his study, including fiber length, micronaire and strength. Myers and Lu [33] reported that micronaire, upper half mean length, fiber strength and elongation were influenced by additive gene action since GCA effects were more significant than SCA effects. Khan et al. [31] agreed with the importance of additive gene action for the major fiber traits. Green and Culp [34], like Myers and Lu [33], showed significant GCA effects for all fiber properties except for the uniformity index. Therefore, early generation selection may be more appropriate for fiber property improvement because effective selection in early generations of segregating material can be achieved when the additive genetic effects are substantial and the environmental effects are low [24,31,35].

Our results are inconsistent with those from Ashokkumar [36] showing GCA variances were lower than SCA variances for all the characteristics he studied, including fiber quality traits. Meredith and Bridge [29] reported non-additive gene action for fiber length. Cheatham et al. [26] reported that fiber fineness and length primarily exhibit dominant genes effects and that fiber elongation is controlled equally by additive and dominant gene effects.

Palve [27] reviewed several studies conducted to estimate the gene action, especially for fiber quality parameters. Inconsistent results were reported, including non-additive [23,37,38], additive and non-additive [6,39], and additive gene action [22,27,40,41]. Contradictions may be due to the different genetic backgrounds of cultivars used, the environmental conditions under which the studies were conducted or the differences in sampling/phenotyping techniques [37].

Cotton from Burkina Faso is exported almost in its entirety, so fiber properties that impact marketing value are referred to as major, i.e., UHML, Str, Mic and +b. For these variables, the best general combiners from this study are TX4, BK2, TX6, TX3 and, to a lesser extent, BK1. These cultivars would be suitable parents for enhancing the major fiber traits in progenies. Similar results with these populations were found in terms of the best parents to enhance agronomic traits. The African lines FK37 (BK1), FK64 (BK2) and E53 (BK6), along with the American lines TX6 and TX4, showed interesting potential for improving components of crop maturity [8]. In Burkina Faso, BK1 has long been the reference control for Str, whereas BK2 is highly appreciated for its exceptional fiber length [42].

The GCA results show that the US lines TX4 and TX6 are potentially good combiners for enhancing Mic, UHML, Str, Elo and Rd (Table 4). Non-domesticated introduced accessions from the USDA NCGC, TX 307 (TX2) and TX 294 (TX1), would result in lower UHML (−3.19 and −0.5, respectively), Str (−2.58 and −1.96, respectively) and UI (−054 and −1.46, respectively) or in increasing SFI (0.77 and 2.20, respectively). Myers and Lu [33], investigating a subset of American cotton cultivars, also concluded that obsolete lines would not contribute positively to a breeding program seeking to develop cultivars with improved fiber traits.

4.2. SCA of Hybrids

The SCA refers to the dominance effects and non-additive interactions of genes, resulting from gene complementation between parents, as dominant or epistatic [4]. In the present study, mean squares analysis revealed that the SCA effects were always low and sometimes not significant (Mic, UHML, Rd) compared with the GCA effects. Therefore, early generation selection could be more effective in fiber quality improvement.

When analyzing the SCA in this study, crosses within American materials exclusively resulted in hybrids expressing significant effects in a contrary direction to be promising for developing new cultivars for Burkina Faso. Traits of lower importance are enhanced, i.e., UI increasing and SFI decreasing, but major properties needing improvement could be decreased, i.e., UHML and Str by −3.53 (TX2xTX4), −2.67 (TX1xTX3) and −2.26 (TX3xTX5) or Elo by −0.48 (TX1xTX3) and −0.44 (TX4xTX6). Previous studies have reported that a narrow genetic base, with parents coming from a common origin or sharing common parents in their pedigree, can lead to insignificant or unfavorable SCA effects [26]. Myers and Lu [33] also reported that American obsolete varieties do not contribute positively to a breeding program focused on developing improved fibers.

From crosses exclusively within Burkina Faso materials, six hybrids exhibited significant SCA effects, implicating mainly BK4 (E9), BK5 (E32) and BK3 (Stam 59A) (in three hybrids each), BK1(FK37) (in two hybrids) and BK6 (E53) (in one hybrid) as being important. This material included both cultivars (BK1 and BK3) and locally collected materials (BK4, BK5 and BK6). BK4 was in more hybrids expressing significant SCA effects for many traits; however, regarding major traits, BK1 was better at transmitting superior genes to its progenies. The hybrid BK5 x BK4 would be interesting, as Mic (−0.63), +b (−0.83) and Rd (+2.31) were potentially enhanced but UHML, Str, UI and SFI could decline. In combination with BK3, BK1 significantly enhanced +b (−0.82) but not the other traits (UHML, Str, UI, SFI and Rd). However, the most appreciable influence of BK1 is perceptible in hybridization with BK4, where UHML (+1.43), UI (+2.30), SFI (−2.44) and Str (+2.48) were significantly enhanced but Mic and Rd were not (Table 5). These parents and cross combinations could be used in future studies to help to achieve cultivars with improvements in major fiber characteristics. It is surprising that BK2 (FK64), the Burkina Faso cultivar standard, does not contribute significantly to progenies even for UHML, whereas the best GCA parents do not always contribute significantly to progenies [38].

Most hybrids (14) exhibiting at least one significant SCA effect were cross combinations between American and Burkina Faso lines. For all these hybrids, American lines were the female parents and Burkina Faso lines were the male parents, with BK4 being the male parent in five hybrids. BK4xTX6 resulted in significant SCA for Mic (+0.55), UHML (−1.82), Str (−2.05), Elo (+1.35) and +b (−1.08). BK4 (E9) contributed both negative trait potentialities and/or inhibited positive effects from the other parent [8,16] because it is a primitive accession not well adapted to commercial cultivation.

Similarly, the older US non-domesticated accessions TX2 and TX1 were shown in four and two cross combinations, respectively, to exhibit at least one significant SCA. In respect to their potential as parents (GCA), the progenies BK1xTX2, BK5xTX2, BK5xTX1 and BK4xTX1 (Table 5) resulted in traits in opposite to desired direction because they inhibited positive effects from the male parent. Similar results were reported for older American cultivars for agronomic traits, even for these same TX2 and TX1 genotypes [8,32].

More modern US experimental lines were also parents in hybrids exhibiting at least one significant SCA. The lines TX5 (16-2-418BB) and TX6 (15-3-416) were in one and two hybrids showing significant SCA, respectively, but none of them (BK4xTX5, BK4xTX6 or BK1xTX6) expressed a combination of desired traits (Table 5). The lines TX3 (16-2-216FQ) and TX4 (15-10-610-7) were in hybrids exhibiting at least one significant SCA effect each, with the hybrids BK6xTX3, BK6xTX4 and BK5xTX4 being potentially interesting as they had significantly improved Str and UI but not Mic, UHML, SFI, Rd or +b. TX3, TX4 and to a lesser extent BK6 were among the best general combiners for almost all the investigated traits. These four parents generally (and three cross combinations specifically) could be used in future studies aimed at achieving cultivars with improvement in fiber characteristics important to Burkina Faso’s cotton commerce.

In summary, in this experiment, the hybrids with desirable SCA effects involved at least one or both high general combiners as parents, which is in concordance with previous studies [3,4,13]. The possibilities to realize value from these populations exist through careful consideration of the differences in germplasm materials, environmental effects, and breeding procedures effective for enhancing fiber traits, including early generation selection in the progenies [3,31,35].

4.3. Heritability

Heritability estimates (h²) were classified as high (>0.50), medium (0.30–0.50) and low (<0.30) according to Bhateria et al. [43]. Narrow-sense heritability was close to broad-sense heritability for all the traits investigated. The two heritability values being nearly equal reflects that there is little difference between the additive and genotypic variance for these traits. Elo was classified as high (0.57), both +b and Str were classified as low (0.17 and 0.09, respectively) and all the other traits exhibited medium heritability, indicating that selection for major marketing traits (+b and Str) appeared less effective than selection for Elo in our study environment.

The heritability values for fiber properties are often reported as high or medium, but low heritability values are also normal [6]. Shao et al. [44], studying NLS cultivars, reported high heritability values from 66.99 (UHML) to 76.67% (Str) for four main fiber properties (UHML, Str, Mic and Elo). From a study on GM cotton by Rehman et al. [45], the UHML, Str and fineness showed high broad-sense heritability estimates, i.e., 52.95%, 59.66% and 70.42%, respectively. Khan and Azhar [46] and Killi et al. [47] found 96% and 94%, whereas Ahmed et al. [48] and Abbas et al. [49] reported 52% and 56% heritability estimates for UHML, respectively; therefore, selection can be useful for improving fiber length. Killi et al. [47], Desalegn et al. [50], Rasheed et al. [51] and Khokhar et al. [52] determined 73%, 33%, 70% and 68% heritability for Str, respectively. Hendawi et al. [53] and Lu et al. [54] estimated 67% and 73% heritability for fiber fineness, respectively.

A range of narrow-sense heritability estimates for fiber quality traits summarized from a divergent sample of genetic populations, various selection units and test instruments is: UHML (0.10–1.00), Str (0.10–0.86), Elo (0.36–0.90) and Mic (0.08–0.53) [55]. This study shows a wide range for these traits globally; the differences might be due to the various experimental materials used with diverse genetic backgrounds [27,35]. Moreover, high estimates of heritability indicate a role for additive gene action in the inheritance of targeted traits and revealed that successful and effective selection can be effectively applied in their improvement. In contrast, low estimates of heritability indicate that the traits were affected by the experimental conditions (small population, environment, etc.) and suggest that sources with high variability for these characteristics should be sought to make improvement.

4.4. Correlation Effects

The significant correlations between the traits investigated were both positive and negative. Favorable significant correlations such as the positive associations between UHML and Str or Rd and between Str and UI or the significant negative correlations between UHML and Mic and between Str and SFI (−0.69) suggested the possibility for the simultaneous improvement of long, strong, fine and high grade fibers; this result is in agreement with those from Liu et al. [34]. Bechere et al. [56] also reported that fiber length had a positive association with fiber strength. Our results were similar to those from Liu et al. [35] regarding undesirable but weak associations between Mic and UI or SFI (0.28 to −0.29) and between Elo and Str, SFI or to a lesser extent UHML (−0.05 to −0.46). However, other authors have reported inconsistent results, finding that UHML was negatively associated with Str [45,57,58].

Other studies, including this one and its companion [8], integrate agronomic and fiber property data to continue the investigation of reported negative associations, such as between lint yield and Str or UHML and between Elo and UHML or Str, or positive ones between lint percentage and fiber fineness related traits, etc., emphasizing the challenges and barriers to achieving both high yield and high fiber quality in cotton cultivars [35,59,60].

5. Conclusions

The cotton fiber produced in Burkina Faso is exported for yarn processing and textile applications, so it is important to consider fiber properties in the breeding process of developing future cultivars. Evaluation of the parents and F₁ hybrids in this study revealed the potential to enhance fiber properties via crosses within materials originating from Burkina Faso by crossing elite cultivars with locally collected germplasm. However, keeping in mind the importance in a breeding program to introduce genetically diverse parental material, long-term future cultivar development could include AgriLife improved breeding lines or other sources with good fiber quality characteristics. Considering both agronomic characteristics and fiber properties, the best parents identified from this study for building on future cultivar development for Burkina Faso include the elite cultivars (FK37 and FK64) × locally collected germplasm (E32 and E53) or × breeding lines (15-10-610-7 and 16-2-216FQ). In addition to identifying recommended parents and cross combinations, analysis of interesting hybrids revealed valuable information toward applying adequate breeding procedures such as selecting for improved fiber in early generations when developing elite cultivars. The ultimate objective of investigation, to develop strategies for developing high value cultivars for the Burkina Faso cotton industry, is ongoing and includes completing the other half of a full diallel, exploring further germplasm exchange opportunities, and investigating all the enhancement values from this opportunity.