Simultaneous Selection of Sweet-Waxy Corn Ideotypes Appealing to Hybrid Seed Producers, Growers, and Consumers in Thailand

Abstract

Multi-trait selection helps breeders identify genotypes that appeal to divergent groups of preferences. In this study, we performed simultaneous selection of sweet-waxy corn hybrids on several traits covering the perspectives of consumers (taller kernel depth, better eating quality), growers (early maturity, shorter plant stature, and high ear yield), and seed producers (high flowering synchrony, acceptable seed yield, and good plant architecture). Three supersweet corn lines and 8 waxy corn lines were intercrossed to generate 48 F1 hybrids according to North Carolina Design II, and these genotypes were laid out in a randomized complete block design with 3 replications across 2 seasons between 2017 and 2018. A sensory blind test on sweetness, stickiness, tenderness, and overall liking was conducted to assess the eating quality of steamed corn samples. Two methods of simultaneous selection, namely unweighted selection index and overall rank-sum index (ORSI), were applied to rank crosses, following all targeted groups of preferences. Genetic parameters and genetic gain were estimated to evaluate the effectiveness of those selection methods. Both approaches had similar patterns of preferable realized gain on each given trait and could identify similar top five crosses with only slight order changes, implying that these methods were effective to rank genotypes according to given selection criteria. One of the tested crosses, 101L/TSC-10 × KV/mon, consistently had the highest unweighted selection index in the dry (7.84) and the rainy (7.15) seasons and the lowest ORSI (310), becoming a promising candidate as synergistic sweet-waxy corn hybrid appealing to consumers, growers, and seed producers. The expected ideotypes of sweet-waxy corn hybrid are discussed.

1. Introduction

Waxy corn (Zea mays L.) is considered a vegetable crop in most Asian countries. In Thailand, it is harvested at the fresh stage and is consumed as a cooked ear [1]. Due to its popularity, further improvement of traditional waxy corn is currently directed on both yield and palatability. Hybrid breeding is most common in corn by exploiting hybrid vigor, well known as heterosis [2]. Moreover, single-cross hybrids account for most corn acreage because of their high yield and uniformity in appearance [3,4]. However, breeding programs for waxy corn are occasionally disconnected from the demands of end-users [5]. An effective strategy to increase cultivar adoption is the development of plant ideotypes considering the divergent preferences of end-users.

Three groups of stakeholders are critically important for vegetable corn breeders: consumers, growers, and seed producers. While traditional waxy corn is recognized for its high stickiness due to a high proportion of amylopectin [6], today’s consumers prefer more palatable corn with balanced flavor, texture, and aroma. Consumers prefer a sweet flavor [7] and a texture with high stickiness and soft tenderness [8]. Farmers may have diverse criteria to select corn cultivars following final product attributes and socioeconomic variables [9]. For farmers in tropical [10], sub-humid [11], and temperate [12,13] zones, cultivars showing early maturity, short plant and ear height, good seed set, and high final yield are appealing. These agronomic attributes fall within waxy corn production in Thailand with a typical tropical savanna climate that often encounters a prolonged dry season and short, extreme rainfall [14]. Finally, seed producers are concerned with parental inbred lines to achieve efficient hybrid seed production. Female inbreds should be vigorous, with low ear placement and high seed yield potential, whereas male inbreds should display abundant pollen production with an extended pollen shedding period, tall plant stature, and good flowering synchrony with female plants [3].

Sweet-waxy corn hybrids have been proposed to improve the eating quality of waxy corn through the concept of synergistic corn and Mendelian ratio between waxy and sweet kernels [1,15]. As a preliminary step, several sweet and waxy corn inbred lines with good general combiners for agronomic traits [16] and sweet-waxy corn hybrids possessing two to three recessive genes with a balanced proportion of total sugar, amylopectin, and phytoglycogen [17] have been identified. The question is whether sweet-waxy corn hybrids appeal to consumers, farmers, and seed producers. Simultaneous selection for multiple traits allows breeders to make their decisions according to selection indices [2]. Weighted selection index is the most effective method for plants [18] if some parameters such as relative economic weights, phenotypic, and genotypic correlations of traits are previously known. Two alternative methods, namely unweighted selection index [19] and overall rank-sum index (ORSI) [20], have been applied to simply rank genotypes based on indices without considering certain economic values of traits of interest. The first mentioned method is commonly used in studies on population improvement, such as in spring wheat [21], whereas the ORSI method is preferable to yield trials [20].

The effectiveness of the selection method could be determined by realizing the genetic gain as the observed differences in average phenotypic value due to selection [22]. Previous studies reported the desired genetic gain on grain yield and yield components through index selection [23,24,25,26], and this approach allows better distribution of desirable gains in the other traits [27]. Since the reliability of simultaneous selection depends upon the selection criteria, genetic material used, and breeding objectives [25], defining the suitable method for multi-traits selection is imperative. However, there is no study reporting the effectivity of simultaneous selection through either selection index or ORSI in hybrid breeding for synergistic sweet-waxy corn.

Therefore, this study aimed (i) to estimate the genetic parameters and genetic gain on attributes representing the three groups’ perspectives of sweet-waxy corn genotypes, (ii) to perform simultaneous selection with unweighted selection index and ORSI on given criteria, and (iii) to investigate whether a single sweet-waxy corn hybrid can meet the preferences of these groups. The information obtained in this study would determine whether and to what extent simultaneous selection based on the given criteria is effective at grouping these hybrids and could be adopted in hybrid breeding for synergistic corn.

2. Materials and Methods

2.1. Plant Materials and Crossing

Eleven parental lines were used in this study. These parental lines (

Table 1. Per se performance of 11 parental lines used in the study.

Parental Lines	Genotype	Source of Ancestors	Per Se Performance
			DTA 1	PHE 1	HYI 1	TSU 2	AMY 2	TSA 2
supersweet corn
101 LBW	bt2bt2Sh2Sh2wx1wx1	USA	53	170.3	6.8 (*)	140.0 (*)	38.1	45.4
101 L/TSC-4	bt2bt2sh2sh2wx1wx1	Thai/USA	47 (*)	119.3 (*)	4.2	108.7	61.1	69.0
101 L/TSC-10	bt2bt2sh2sh2wx1wx1	Thai/USA	46 (*)	102.3 (*)	4.9	87.2 (*)	62.4	68.5
waxy corn
YINNUO 18	Bt2Bt2Sh2Sh2wx1wx1	China	58	149.3	6.9 (*)	33.8	150.7	157.5
CAITIANNUO 13-1	Bt2Bt2Sh2Sh2wx1wx1	China	55	144.2 (*)	8.1 (*)	39.6	136.6	145.7
HONGYU 2	Bt2Bt2Sh2Sh2wx1wx1	China	54	138.4 (*)	7.3	20.3	143.4	151.7
HJ	Bt2Bt2Sh2Sh2wx1wx1	China	61	143.0	5.2	39.2	105.8	121.9
ORANGE WAXY 13	Bt2Bt2Sh2Sh2wx1wx1	China	57	167.6	5.0	32.1	126.3	132.4
KV/mon	Bt2Bt2Sh2Sh2wx1wx1	Thai/USA/Vietnam	49 (*)	140.5 (*)	6.9	33.1	177.5 (*)	181.6 (*)
KV/3473	Bt2Bt2Sh2Sh2wx1wx1	Thai/USA/Vietnam	52 (*)	125.1 (*)	4.7	30.2	153.7	165.6
KNM102	Bt2Bt2Sh2Sh2wx1wx1	Thailand	54	129.9	2.2	104.7	32.7	122.2

DTA anthesis date (days after sowing). PHE plant height (cm). HYI husked yield (ton ha⁻¹). TSU total sugar (mg g⁻¹). AMY amylopectin (mg g⁻¹). TSA total starch (mg g⁻¹). ¹ Dermail et al. [16]. ² Fuengtee et al. [17]. (*) good general combiners regarding GCA and per se. bt2 brittle2. sh2 shrunken2. wx1 waxy1.

) consisted of three and eight lines of supersweet corn and waxy corn, respectively. Genotype 101LBW extracted from ATS-2, a supersweet corn hybrid cultivar in Thailand carrying bt2 mutant, is double recessive (bt2 and wx1), with moderate maturity, tall stature, and high yield and sugar content. Genotypes 101L/TSC-4 and 101L/TSC-10 are triple recessive (bt2, sh2, and wx1) with early maturity, short stature, and moderate yield and sugar content. These inbred lines are derived from the same ancestors: supersweet corn line 101bt (bt2bt2 Sh2Sh2 wx1wx1) and temperate supersweet corn hybrid TSC (Bt2Bt2 sh2sh2 Wx1Wx1). Eight waxy corn lines recessive for the wx1 gene have diverse maturities (early, moderate, and late) and are adapted to different climatic zones (tropical Thailand and subtropical China). All tested waxy corn lines combine moderate to high amylopectin with low sugar content.

To generate synergistic sweet-waxy corn hybrids, 3 sweet corn lines were designated as group I and 8 waxy corn lines as group II to generate 48 F1 hybrids (Figure 1), including reciprocals, by following the North Carolina Design II [28]. Those hybrids were produced at the Vegetable Experimental Farm, Khon Kaen University, Thailand, in 2017. Due to the different maturity levels of our parental lines, staggered planting of sweet and waxy corn lines was conducted two and three times, respectively, to ensure pollination.

2.2. Experimental Design

Eleven parental lines and 48 F1 progenies were laid out in a randomized complete block design (RCBD) with 3 replications and evaluated in the dry season of 2017/2018 and the rainy season of 2018 at the Agronomy Field Crop Station, Khon Kaen University, Thailand (16°28′27.7″ N, 102°48′36.5″ E; 190 masl). The parental lines and hybrids were planted in adjacent blocks in the same field; therefore, these blocks were separately randomized within each replicate. This modification was performed to avoid several drawbacks such as border, shading, and competition effects. Each plot consisted of 2 rows of 5 m length with 75 cm and 25 cm row and plant spacing, respectively. The crop field management followed the Department of Agriculture, Thailand recommendations [29], including fertilization, irrigation, and pest, disease, and weed control.

2.3. Data Collection

The anthesis date (days after sowing/DAS) was determined as the number of days from sowing to when 50% of the plants shed pollen. The silking date (DAS) was determined as the number of days from sowing until silks emerged on 50% of the plants. The anthesis-silking interval (ASI) of each pairwise combination was calculated as the silking date of a female parent minus the anthesis date of a male parent to obtain the information on flowering synchrony.

After the milk stage, 10 random plants (5 from each row in a plot), excluding border plants, were recorded for plant height (cm) as a distance from ground level to the node bearing the flag leaf, and ear height (cm) as a distance from ground level to the node bearing the uppermost ear. The plant height difference (cm) of each pairwise combination was calculated as plant height of the male parent minus plant height of the female parent. Ear position was calculated as the ratio between ear and plant height of female parents.

Husked ear yield, husked ear length (cm), kernel depth (cm), row number per ear, and kernel number per row were measured after harvest at the fresh stage (21 days after open pollination/DAP). The first DAP was counted when the corn silk emerged to at least 2.5 cm. The total kernel number per ear was calculated by multiplying row number per ear by kernel number per row. Husked ear yield was based on averaged best 10 ears per plot and then converted into ton ha⁻¹.

2.4. Sample Preparation and Sensory Evaluation

Ten representative corn ears with physiologically undamaged and uniform ears harvested at the fresh stage (21 DAP) were used as a plot sample. These samples were derived from sibling pollination to keep genetic purity from any unintended pollen contamination. Sample preparation followed Simonne et al. [20] and Harakotr et al. [30] with proper modification. The whole ear was husked, and 3 cm of both ear tip and base were removed to reduce kernel maturity variation. Steaming, a common method of cooking fresh corn, was chosen in this study. Husked ears were steamed in a single layer in a covered stainless steel steamer for 45 min, cooled in iced water for 1 min, packed in a plastic bag, and stored in a freezer at −20 °C until the day of the sensory evaluation. Although all corn genotypes possessed varied maturity levels and were harvested at different dates, this procedure allowed all samples to be treated equally by immediately steaming the fresh corn ears on the day of harvest of each genotype and directly freezing the steamed corn ears to minimize significant quality degradations and to retain similar experimental error.

A modified sensory blind test was performed on frozen samples following Simla et al. [15]. On the day of the sensory test, both frozen samples and check ears were heated in steaming water for 30 min and then cut into 3 cm pieces and placed on a dish. Fifteen panelists represented different age groups, familiarity levels with the eating quality of corn, and gender. These panelists were trained for the parameters of eating quality and the scaling procedure in a 10 min session and were asked to taste, evaluate, and rate the sample for 4 eating quality attributes, namely sweetness, stickiness, tenderness, and overall liking. Fourteen rating scales were applied for these parameters. Two commercial checks, namely supersweet corn hybrid Hi-brix 3 from Pacific Seeds (Thai) Co., Ltd., Saraburi, Thailand and waxy corn hybrid Niyw Mwang Taem (NMT) from Syngenta Seeds (Thai) Co., Ltd., Bangkok, Thailand were included as parameter standards of sweetness and stickiness. These checks were derived from a traditional market as fresh, unhusked ears. For sweetness, rating scales ranged from 0 (not sweet) to 14 (extremely sweet), represented by waxy corn hybrid Niyw Mwang Taem (NMT) and supersweet corn hybrid Hi-brix 3, respectively. For stickiness, rating scales from 0 (not sticky) to 14 (extremely sticky) were represented by supersweet corn hybrid Hi-brix 3 and waxy corn hybrid Niyw Mwang Taem (NMT), respectively. For tenderness, the panelists were asked to chew the samples 4 to 5 times per sample and to rate the samples from 0 (no creamy texture) to 14 (highly creamy texture). Overall liking was scored with ratings from 0 (most unfavorable) to 14 (most favorable).

2.5. Data Analysis

Two-season data for each observed trait were subjected to combined analysis of variance (ANOVA) in RCBD using PROC MIXED of SAS ver. 9.0 [31] using the following linear model:

$${\mathrm{Y}}_{\mathrm{sbhr}}{=\mu +\mathsf{\alpha }}_{\mathrm{s}}{+\mathsf{\beta }}_{\mathrm{b}}\left({\mathsf{\alpha }}_{\mathrm{s}}\right){+\mathsf{\gamma }}_{\mathrm{h}}{+\mathsf{\delta }}_{\mathrm{r}}{+\mathsf{\gamma }}_{\mathrm{h}}{\mathsf{\delta }}_{\mathrm{r}}{+\mathsf{\alpha }}_{\mathrm{s}}{\mathsf{\gamma }}_{\mathrm{h}}{+\mathsf{\alpha }}_{\mathrm{s}}{\mathsf{\delta }}_{\mathrm{r}}{+\mathsf{\alpha }}_{\mathrm{s}}{\mathsf{\gamma }}_{\mathrm{h}}{\mathsf{\delta }}_{\mathrm{r}}{+\mathsf{\epsilon }}_{\mathrm{sbhr}}$$

(1)

where s = 1, 2; b = 1, 2, 3; h = 1, 2, 3…24; r = 1, 2, 3…24; Y_sbhr denotes the phenotype of hybrid h and reciprocal r in season s and block b; µ is the overall mean; α_s is the effect of season s; β_b(α_s) is the effect of block b nested within season s; γ_h is the effect of hybrid h; δ_r is the effect of reciprocal r; γ_hδ_r is the effect of the orthogonal linear contrast between hybrid h and reciprocal r; α_sγ_h is the effect of the interaction between season s and hybrid h; α_sδ_r is the effect of the interaction between season s and reciprocal r; α_sγ_hδ_r is the effect of the interaction between season s and orthogonal contrast between hybrid h and reciprocal r; ε_sbhr is the pooled error of hybrid h and reciprocal r in season s and block b.

Tukey’s range test (HSD) at p ≤ 0.05 was used for mean comparison between each selected cross mean and the grand mean [32]. Selection criteria were highlighted in three groups of attributes according to their preferences, namely seed producer, grower, and consumer. The first group, seed producer’s perspective, focused on parental pair performance for good flowering synchrony (ASI, plant height difference, ear position) and kernel set (husked ear length, total kernel number per ear). The second group, the grower’s perspective, focused on F1 hybrid means for early maturity (anthesis date, silking date), lodging tolerance (plant height, ear height), and high yield (husked ear yield). The third group, consumer preference, focused on F1 hybrid means for better eating quality (kernel depth, sweetness, stickiness, tenderness, overall liking). All tested crosses were then ranked by using simultaneous selection through the unweighted selection index [33] and overall-rank sum index (ORSI) [20].

For the unweighted selection index, each tested cross mean was standardized prior to calculating the selection index. The standardized cross mean (p_ij) was calculated as follows:

$${\mathrm{p}}_{\mathrm{ij}}=\frac{{\mathrm{x}}_{\mathrm{ij}}{ - \mathrm{m}}_{\mathrm{i}}}{{\mathrm{s}}_{\mathrm{i}}}$$

(2)

where x_ij is mean of each cross; m_i grand mean of each trait; s_i standard deviation. Index for hybrid selection was determined based on the following formula:

$$\mathrm{Z}=-{1\mathrm{p}}_1-{ 1\mathrm{p}}_2{+1\mathrm{p}}_3-{ 1\mathrm{p}}_4-{ 1\mathrm{p}}_5-{ 1\mathrm{p}}_6{+1\mathrm{p}}_{7 }{+1\mathrm{p}}_8{+1\mathrm{p}}_9$$

(3)

where Z is selection index; p₁ is the anthesis-silking interval of pairwise combination; p₂ is ear position of female; p₃ is total kernel number per ear of female; p₄ is plant height difference of pairwise combination; p₅ is anthesis date of hybrid; p₆ is plant height of hybrid; p₇ is husked ear yield of hybrid; p₈ is kernel depth of hybrid; p₉ is overall liking of hybrid.

For ORSI, each cross was ranked for each attribute within a group. The top cross was assigned rank 1, in which all rankings were oriented the same way regarding desirability [34]. Then, the overall rank-sum (ORS) for each entry for each group was attained by accumulating the ranks of all attributes within a group and was visualized with a dendrogram based on hierarchical Ward’s clustering method by JMP Pro software [35]. The global index or ORSI of each cross was calculated by adding up the ranks across groups and seasons [20]. Therefore, a lower ORSI represented a promising sweet-waxy corn hybrid.

Realized genetic gain (%) (RG%) was estimated for accessing the effectivity of simultaneous selection through 2 methods: selection index and ORSI. The RG (%) was calculated relative to the population mean of all 48 hybrids, following formula proposed by Singh and Chaudhary [36] and Souza et al. [37] as:

$$\mathrm{RG} (\%)=\frac{\mathrm{MS} - \mathrm{MP}}{\mathrm{MP}} \times 100$$

(4)

where MS is mean of top five hybrids derived from simultaneous selection and MP is population mean of 48 hybrids.

Genetic parameters were estimated from the expected mean squares [38]. Broad-sense heritability was estimated using the variance ratio [39] as follows:

$${\mathrm{h}}_{\mathrm{bs}}^2=\frac{{\mathsf{\sigma }}_{\mathrm{g}}^2}{{\mathsf{\sigma }}_{\mathrm{g}}^2{+(\mathsf{\sigma }}_{\mathrm{gs}}^2{/\mathrm{s})+(\mathsf{\sigma }}^2/\mathrm{rs})}$$

(5)

where ${\mathrm{h}}_{\mathrm{bs}}^2$ is broad-sense heritability estimates; ${\mathsf{\sigma }}_{\mathrm{g}}^2$ is genotypic variance; ${\mathsf{\sigma }}_{\mathrm{gs}}^2$ is variance of the interaction between season and genotype; ${\mathsf{\sigma }}^2$ is variance of error; s is the number of seasons; and r is the number of replications.

In each growing season, the top 5 from 48 hybrids were respectively selected through each selection method; thus, the selection intensity was about 10%. Genetic advance (GA) as predicted genetic gain was calculated following Singh and Chaudhary [36] formula:

$$\mathrm{GA}=\mathrm{i} \times {\mathsf{\sigma }}_{\mathrm{p}} \times {\mathrm{h}}_{\mathrm{bs}}^2$$

(6)

where $\mathrm{i}$ is selection intensity, which is 1.76 at 10% and ${\mathsf{\sigma }}_{\mathrm{p}}$ is phenotypic standard deviation.

3. Results

3.1. Combined Analysis of Variance (ANOVA)

Mean squares of each source of variation for seed producer preferences are presented in

Table 2. Combined ANOVA for seed producer preferences including ASI (anthesis-silking interval), PHD (plant height difference), EP (ear position), HEL (husked ear length), NRE (row number per ear), NKR (kernel number per row), and TKE (total kernel number per ear) of 48 sweet-waxy corn crosses.

Source	df	Mean Squares
		ASI	PHD	EP 1	HEL	NRE	NKR	TKE
Season	1	61.4 *	4.7 2 ns	1.8 ns	940.6 **	121.2 *	3987.9 **	1016334.0 **
Block/season	4	5.2 **	0.0 ns	0.2 **	6.6 *	5.9 **	43.4 **	13652.2 **
Cross	47	298.9 **	6637.0 **	4.6 **	16.0 **	12.2 **	69.2 **	26897.9 **
Reciprocal (rec)	23	119.4 **	6081.8 **	2.7 **	3.9 **	10.5 **	1.7 ns	3564.6 **
Hybrid	23	106.3 **	6081.8 **	3.4 **	26.0 **	13.2 **	125.6 **	44833.4 **
Hybrid vs. rec	1	8855.5 **	32173.3 *	75.8 ns	61.5 ns	29.3 ns	326.6 ns	151048.4 ns
Season x cross	47	12.4 **	284.1 **	0.1 **	7.9 **	0.9 **	24.5 **	5797.7 **
Season x rec	23	11.7 **	285.8 **	0.1 *	8.9 **	0.9 **	28.3 **	5957.8 **
Season x hybrid	23	10.8 **	285.8 **	0.1 *	2.4 ns	0.7 **	13.6 **	4356.0 **
Season x (hybrid vs. rec)	1	63.2 **	206.0 ns	0.1 ns	109.2 **	3.9 **	187.3 **	35276.1 **
Pooled error	188	1.2	95.0	0.1	1.9	0.3	6.2	1012.9
CV (%)		12.95	1.18	5.55	11.27	4.48	11.33	11.29

CV coefficient of variation. ** and * significant at p ≤ 0.01 and p ≤ 0.05, respectively. ns not significant at p ≥ 0.05. ¹ value in 10⁻². ² value in 10⁻²⁶.

, while the grower and consumer preferences are in

Table 3. Combined ANOVA for grower and consumer preferences including AD (anthesis date), SD (silking date), PH (plant height), EH (ear height), HEY (husked ear yield), KD (kernel depth), STC (stickiness), SWT (sweetness), TND (tenderness), and OL (overall liking) of 48 sweet-waxy corn crosses.

Source	df	Mean Squares
		AD	SD	PH	EH	HEY	KD 1	STC	SWT	TND	OL
Season	1	483.0 **	70.0 **	161.8 ns	87.3 ns	326.1 **	82.0 **	4.0 ns	31.3 ns	0.2 ns	35.2 ns
Block/season	4	0.6 ns	0.7 ns	185.6 *	69.7 ns	0.1 ns	2.5 **	49.2 **	75.3 **	49.7 **	52.0 **
Cross	47	34.8 **	35.0 **	2019.1 **	1269.3 **	11.7 **	1.0 **	3.2 **	3.9 **	3.0 **	3.1 **
Reciprocal (rec)	23	35.1 **	36.1 **	1804.5 **	1186.3 **	12.2 **	1.0 **	4.2 **	3.4 *	3.7 **	3.0 **
Hybrid	23	35.0 **	28.9 **	2309.9 **	1403.4 **	11.7 **	1.0 **	2.1 *	4.4 **	2.0 ns	3.4 **
Hybrid vs. rec	1	23.9 ns	150.2 ns	266.6 ns	92.0 ns	0.3 ns	0.1 ns	5.2 ns	5.4 ns	9.2 ns	0.4 ns
Season x cross	47	2.3 **	3.8 **	174.4 **	48.1 ns	1.6 **	0.4 *	3.5 **	3.1 **	2.6 **	2.6 **
Season x rec	23	1.3 ns	2.7 **	128.5 *	28.5 ns	0.9 **	0.2 ns	4.0 **	3.5 *	3.2 **	2.9 **
Season x hybrid	23	1.6 ns	2.3 *	110.5 ns	61.6 *	1.7 **	0.6 **	3.0 **	2.9 ns	2.1 ns	2.5 *
Season x (hybrid vs. rec)	1	39.7 **	66.1 **	2698.0 **	188.8 *	15.23 **	0.9 ns	0.9 ns	0.1 ns	0.8 ns	0.4 ns
Pooled error	188	1.4	1.4	74.3	36.8	0.3	0.2	1.3	1.8	1.4	1.3
CV (%)		2.56	2.57	4.95	7.48	5.28	5.56	12.82	19.85	13.62	13.02

CV coefficient of variation. ** and * significant at p ≤ 0.01 and p ≤ 0.05, respectively. ns not significant at p ≥ 0.05. ¹ value in 10⁻².

. Season was significant for the anthesis-silking interval (ASI), husked ear length (HEL), row number per ear (NRE), kernel number per row (NKR), total kernel number per ear (TKE), anthesis date (AD), silking date (SD), husked ear yield (HEY), and kernel depth (KD). Block was significant for all observed traits except for plant height difference (PHD), AD, SD, ear height (EH), and HEY. Cross was significant for all observed traits. The effect of cross was then partitioned into reciprocal, hybrid, and hybrid vs. reciprocal effects. Reciprocal was significant for all observed traits except for NKR. Hybrid was significant for all observed traits except for tenderness (TND). Hybrid vs. reciprocal was significant for ASI and PHD only.

Interaction between season and cross was significant for all observed traits except for EH. This interaction was then partitioned into season × reciprocal, season × hybrid, and season × (hybrid vs. reciprocal) effects. Season × reciprocal effect was significant for all observed traits except for AD, EH, and KD. Season x hybrid effect was significant for all observed traits except for HEL, AD, PH, sweetness (SWT), and TND. Season × (hybrid vs. reciprocal) effect was significant for most observed traits, including ASI, HEL, NRE, NKR, TKE, AD, SD, PH, EH, and HEY.

3.2. Genetic Parameters and Genetic Gains

Genotypic variation was high on TKE and PHD, moderate on ASI, PH, and EH, and low for the remaining observed traits (

Table 4. Estimates of genetic parameters and genetic gains of seed producer, grower, and consumer preferences at 10% simultaneous selection.

Traits	${\mathsf{\sigma }}_{\mathbf{g}}^2$	${\mathsf{\sigma }}_{\mathbf{p}}^2$	${\mathbf{h}}_{\mathbf{b}\mathbf{s}}^2$	GA	RG 1 (%)	RG 2 (%)
Seed producer preference
Anthesis-silking interval	44.02	45.24	0.97	1151.92	−42.04	−45.89
Plant height difference	995.78	1090.78	0.91	5306.50	−39.46	−72.09
Ear position	0.75	0.85	0.88	143.17	−16.95	−18.77
Husked ear length	0.65	2.55	0.25	71.64	8.33	14.71
Row number per ear	1.68	1.98	0.85	210.37	6.05	6.94
Kernel number per row	1.35	7.55	0.18	86.47	6.73	16.24
Total kernel number per ear	1921.77	2934.67	0.65	6243.58	12.26	22.93
Grower and consumer preferences
Anthesis date	5.12	6.52	0.79	352.77	−5.09	−6.33
Silking date	4.40	5.80	0.76	321.55	−5.64	−6.35
Plant height	274.08	348.38	0.79	2584.44	−9.41	−10.11
Ear height	199.77	236.57	0.84	2285.91	−17.44	−21.77
Husked ear yield	1.90	2.20	0.86	225.45	10.76	14.56
Kernel depth	0.03	0.23	0.14	12.15	4.35	3.01
Stickiness	0.32	1.62	0.19	43.83	0.11	0.07
Sweetness	0.35	2.15	0.16	42.01	11.11	2.06
Tenderness	0.27	1.67	0.16	36.35	6.12	2.61
Overall liking	0.30	1.60	0.19	41.74	5.96	0.68

${\mathsf{\sigma }}_{\mathrm{g}}^2$ genotypic variance. ${\mathsf{\sigma }}_{\mathrm{p}}^2$ phenotypic variance. ${\mathrm{h}}_{\mathrm{bs}}^2$ broad-sense heritability. GA genetic advance. RG1 (%) realized genetic gain based on unweighted selection index. RG2 (%) realized genetic gain based on ORSI method.

). Heritability estimates were high, ranging from 0.84 to 0.97 on ASI, PHD, EP, NRE, EH, and HEY, moderate, ranging from 0.65 to 0.79 on TKE, AD, SD, and PH, and low, ranging from 0.14 to 0.25 on HEL, NKR, KD, STC, SWT, TND, and OL. Estimates of genetic advance (GA) were high on ASI, PHD, TKE, PH, and EH, ranging from 1151.92 to 5306.50, moderate on RNE, AD, SD, and HEY ranging from 143.17 to 352.77, and low on HEL, NKR, KD, STC, SWT, TND, and OL ranging from 12.15 to 86.47.

The realized genetic gain based on unweighted selection index (RG1) was negative on ASI (−45.89%), PHD (−39.46%), EP (−16.95%), AD (−5.09%), SD (−5.64%), PH (−9.41%), and EH (−17.44%) and positive on HEL (8.33%), NRE (6.05%), NKR (6.73%), TKE (12.26%), HEY (10.76%), KD (4.35%), STC (0.11%), SWT (11.11%), TND (6.12%), and OL (5.96%). A similar pattern of gain was noticed for simultaneous selection based on the ORSI method (RG2). The estimates of RG2 was negative on ASI (−45.89%), PHD (−72.09%), EP (−18.77%), AD (−6.33%), SD (−6.35%), PH (−10.11%), and EH (−21.77%) and positive on HEL (14.71%), NRE (6.94%), NKR (16.24%), TKE (22.93%), HEY (14.56%), KD (3.01%), STC (0.07%), SWT (2.06%), TND (2.61%), and OL (0.68%).

3.3. Simultaneous Selection

The top 5 crosses with the highest unweighted selection index (Z) evaluated in the dry season of 2017/2018 were 101L/TSC-10 × KV/mon (7.84), KV/mon × 101L/TSC-10 (6.69), KV/3473 × 101LBW (6.62), 101L/TSC-10 × CAITIANNUO 13-1 (5.84), and 101L/TSC-10 × YINNUO 18 (5.01). Meanwhile, the top five crosses with the highest Z evaluated in the rainy season 2018 were 101L/TSC-10 × KV/mon (7.15), 101L/TSC-10 × KV/3473 (7.13), KV/mon × 101L/TSC-4 (6.40), KV/mon × 101L/TSC-10 (5.76), and 101L/TSC-10 × CAITIANNUO 13-1 (5.49) (

Table 5. Top 5 crosses based on unweighted selection index among 48 sweet-waxy corn crosses evaluated in the dry season 2017/2018 (upper table) and in the rainy season 2018 (lower table).

Cross	Parent	Female		Male	F1 Progeny					Z
	ASI	EP	TKE	PHD	AD	PH	KD	HEY	OL
Dry season
101L/TSC-10 × KV/mon	1	0.33	307	42.0	44	151.3	1.04	11.2	11.7	7.84
KV/mon × 101L/TSC-10	2	0.38	471	−42.0	43	153.4	1.03	12.1	10.7	6.69
KV/3473 × 101LBW	2	0.45	341	52.0	47	190.4	1.20	13.3	10.5	6.62
101L/TSC-10 × CAITIANNUO 13-1	7	0.33	307	52.0	46	150.2	1.07	12.8	9.9	5.84
101L/TSC-10 × YINNUO 18	10	0.33	307	52.5	48	146.7	1.11	12.4	9.8	5.01
Grand mean 1	7	0.42	341	0.0	48	173.5	1.03	11.5	9.3
HSD 5% 1	2	0.03	62	17.2	1	15.9	0.11	0.3	1.8
Rainy season
101L/TSC-10 × KV/mon	0	0.32	225	28.7	42	162.2	0.97	10.9	7.8	7.15
101L/TSC-10 × KV/3473	2	0.32	225	23.6	42	150.8	0.97	10.5	8.2	7.13
KV/mon × 101L/TSC-4	3	0.41	346	−12.6	42	158.2	0.90	11.6	9.5	6.40
KV/mon × 101L/TSC-10	4	0.41	346	−28.7	42	150.5	0.95	10.7	8.7	5.76
101L/TSC-10 × CAITIANNUO 13-1	5	0.32	225	26.6	43	164.8	0.96	10.4	8.7	5.49
Grand mean 2	6	0.44	223	0.0	45	175.0	0.92	9.43	8.6
HSD 5% 2	2	0.04	39	14.3	2	11.7	0.05	0.4	1.9

Z unweighted selection index. ASI anthesis silking interval (days). EP ear position. TKE total kernel number per ear (seeds per ear). PHD plant height difference (cm). AD anthesis date (DAS). PH plant height (cm). HEY husked ear yield (ton ha⁻¹). KD kernel depth (cm). OL overall liking. HSD 5% as critical value of mean comparison between individual top five crosses and the grand mean. ¹ Both grand mean and HSD 5% were derived from 48 crosses in the dry season 2017/2018. ² Both grand mean and HSD 5% were derived from 48 crosses in the rainy season 2018.

).

A variability of sweet-waxy corn hybrids in response to three groups of preferences was revealed in dendrogram based on each overall rank-sum (ORS) in the dry season (Figure 2). Four major groups (A-D) were noticed. For illustration, most hybrids belonging to group A, such as Hongyu2 × 101LBW and 101LBW × Yinnuo18, appealed to the preferences of seed producers and consumers. Some hybrids belonging to group B, such as 101L/TSC-4 × KV/3473 and 101L/TSC-10 × KNM102, appealed to the preferences of seed producers and growers, while other members of this group failed to appeal to our targeted preferences. Conversely, hybrids in group C, such as Hongyu2 × 101L/TSC-10 and KNM102 × 101L/TSC-10, met growers’ demands only. Most hybrids in group D, such as 101L/TSC-10 × KV/mon and its reciprocal KV/mon × 101L/TSC-10, obviously appealed to all target groups. Therefore, the hybrids belonging to this group are potential sweet-waxy corn hybrids to be promoted. When performing simultaneous selection with overall rank-sum index (ORSI), we noticed that the top five crosses with the lowest ORSI were 101L/TSC-10 × KV/mon (360.0), KV/mon × 101L/TSC-10 (412.5), 101L/TSC-10 × CAITIANNUO 13-1 (441.5), KV/mon × 101L/TSC-4 (450.0), and 101L/TSC-10 × KV/3473 (460.5) (

Table 6. Overall rank-sum index (ORSI) of top 5 crosses among 48 possible crosses.

Cross	Overall Rank Sum (ORS)						ORSI 4
	Seed Producer 1		Grower 2		Consumer 3
	Dry	Rainy	Dry	Rainy	Dry	Rainy
101LTSC-10 × KV/mon	45.5	35.0	47.5	58.0	32.0	142.0	360.0
KV/mon × 101L/TSC-10	76.0	85.5	46.0	45.0	34.0	126.0	412.5
101LTSC-10 × CAITIANNUO 13-1	63.5	59.5	48.5	48.5	115.0	106.5	441.5
KV/mon × 101L/TSC-4	65.0	70.0	81.5	67.0	112.0	54.5	450.0
101LTSC-10 × KV3473	67.0	52.0	66.5	36.0	97.5	141.5	460.5
Minimum	45.5	35.0	46.0	36.0	32.0	50.0	360.0
Maximum	160.5	146.0	199.5	200.0	222.0	210.0	894.0

¹ Seed producer includes anthesis silking interval, ear position, total kernel number per ear, and plant height difference. ² Grower includes anthesis date, silking date, plant height, ear height, and husked ear yield. ³ Consumer includes kernel depth, sweetness, stickiness, tenderness, and overall liking. ⁴ ORSI calculated by adding the ranks of each ORS.

).

4. Discussion

4.1. Combined Analysis of Variance (ANOVA)

Both significant effects of season and season x cross interaction indicated that different climate profiles between dry and rainy seasons altered parents’ performances on flowering synchrony and seed set and F1 progenies on agronomic traits and eating quality. Previous studies reported the greater impacts of growing seasons on flowering dates, plant architecture, yield components, and ear yields of purple waxy corn populations [40,41] and small-ear waxy corn accessions [42]. Fuengtee et al. [17] noticed the predominant effect of the growing season on kernel carbohydrate components, including total starch, amylopectin content, total sugar, and sugar fractions. Considering the same genotype samples used in both Fuengtee et al. [17] and our present study, it was assumed that the changes in kernel carbohydrates due to seasonal variations were one plausible factor altering the eating quality of sweet-waxy corn hybrids. Besides, among climate profiles, two parameters, namely temperature and solar radiation, were responsible for the changes of total starch and amylopectin content of immature waxy corn [43,44].

The significant effect of the block on eating quality implied that panelist profiles such as age, gender, and familiarity with traditionally cooked waxy corn became a critical factor on the sensory blind test. These factors have been noticed for driving consumer preferences in deep-fried peanuts [45] and wine [46]. The significant effect of cross for all observed traits indicated that genetic variation among tested crosses on traits of interest existed, and simultaneous selection among 48 sweet-waxy corn F1 crosses can be carried out.

The reciprocal cross effect results from the accumulation of doses of both maternal and non-maternal effects [47]. In this study, hybrid vs. reciprocal represented the reciprocal cross effect. The significance of this effect on ASI and PHD indicated that parental pair selection for hybrid seed production on flowering synchrony was critical. This result corroborated a classic breeding rule that “female plant should be a female, male plant should be a male”. Therefore, performing a full-sib mating scheme followed by careful selection of parental pairs may improve the flowering synchrony and kernel yield, benefiting hybrid seed producers. In addition, a lack of significance of reciprocal cross effect on eating quality suggested that vegetable corn breeders have flexibilities in assigning whether sweet or waxy corn is a female plant or vice versa.

4.2. Genetic Gains Reflect the Effectiveness of Simultaneous Selection

Improvements on multiple traits rather than a single trait are common in current crop breeding worldwide to ensure the high adoption of the cultivars. That breeding goal could be achieved through simultaneous selection. Variance components and genetic gains are routinely estimated to evaluate the effectiveness of the selection method applied to certain breeding objectives [48]. Previous studies reported larger genetic gains from simultaneous selection than single-trait selection in several crops such as rice [49], sugarcane [50], wheat [51], and maize [26]. In our study, the realized gain of both ORSI and selection index was assumed as the percent average differences between selected means and grand means. The negative realized gain on ASI, PHD, EP, AD, SD, PH, and EH but positive estimates on HEL, NRE, NKR, TKE, HEY, KD, STC, SWT, TND, and OVL indicated two things: (1) simultaneous selection performed was able to distribute the desired gains on each targeted trait and (2) the preferred ideotypes could be obtained, such as good flowering synchrony and high female seed set for efficient hybrid seed production, early flowering date, shorter plant stature, higher ear yield, and better eating quality of the hybrids. Vieira et al. [52] performed selection indices in popcorn composites, and high and balanced selection gains on popping expansion and grain yield were obtained. Besides, Crispim-Filho et al. [26] applied several multiple-trait strategies in maize to obtain more balanced predicted gains on early flowering, short flowering interval, plant architecture, and yield components.

Simultaneous selection with weighted selection indices using the Smith-Hazel index has been applied to improve biomass yield and quality attributes of switchgrass [53] and grain yield of wheat genotypes [54]. This selection index, which requires the assignment of the relative economic values, is ideal from a theoretical point of view; however, it is quite challenging from a practical point of view [55]. Unstable relative economic values over time and locations, traits dependence composing multiple regression equations, and lack of a standard to assign the economic values for some traits were reasons to explain the limitation of weighted selection indices [56,57]. For practical purposes, index selection could be calculated by considering desired gains on important traits according to the breeder without assigning the relative economic weights to each trait [58]. An unweighted selection index as an alternative method has been adopted, and it was effective in selection studies of early maturing, disease-resistant spring wheat genotypes [21] and early soybean inbred lines [59]. In our study, we obtained preferable realized genetic gain on each selection criterion and slight order changes of the top five crosses derived from the unweighted selection index and ORSI, indicating that both approaches were effective in simultaneous selection. The first method mentioned above allowed breeders to change the sign of each variable in an equation according to the nature of traits and aim of selection. This helps breeders in automatic ranking, especially when there are many traits to evaluate and numerous genotypes to rank. On the contrary, ORSI is a simple method for multi-trait selection with manual ranking of tested genotypes for each trait. This method was suitable for evaluation studies with few genotypes. However, it was more time-consuming and less efficient with increasing numbers of genotypes. Previous studies performed simultaneous selection based on ORSI among limited numbers of genotypes, such as 10 sweet corn varieties [20] and 6 muskmelon varieties [60].

4.3. Candidates of Sweet-Waxy Corn Hybrids and the Expected Ideotypes for Three Groups of Preferences

The seasonal effect significantly existed on four selection criteria: ASI, TKE, AD, HEY, and KD in simultaneous selection based on unweighted selection index. Accordingly, the top five crosses were shifted between two seasons. This result illustrated the sensitivity of the unweighted selection index to rank the genotypes over different weather conditions, and it is helpful for multi-environmental field trials. Our tested hybrid 101L/TSC-10 × KV/mon consistently had the highest unweighted selection index in the dry and the rainy seasons and the lowest ORSI. Hence, this cross was the candidate for sweet-waxy corn hybrids appealing to consumers, growers, and seed producers. This hybrid came from parents with good per se and general combining ability (GCA). Supersweet corn line 101L/TSC-10 performed good per se and GCA for anthesis date, plant height, and total sugar, whereas the male KV/mon for anthesis date, plant height, husked ear yield, and amylopectin [16,17]. This finding illustrates that each parent with good per se and GCA could share favorable alleles to the progeny. A previous study using tropical shrunken sweet corn lines reported that mid-parent values and GCA effects significantly contributed to hybrid and parent performances for some important agronomic traits, yield components, and kernel qualities [61].

Good parent ideotypes of 101L/TSC-10 × KV/mon such as high flowering synchrony, moderate seed yield, and lower ear position of female 101L/TSC-10 and higher plant stature of the male KV/mon may assist seed producers in efficient hybrid seed production by means of reduced cost of goods sold (COGS) and acceptable market price of seed [4], leading to higher farmer’s willingness to pay [62]. In addition, the impressive hybrid performance of 101L/TSC-10 x KV/mon, like early maturity, shorter plant height, and higher husked ear yield, is preferred by growers in the tropical zone that require corn ideotypes with lodging and drought tolerances and high marketable yield [10]. Growers in Thailand can cultivate this early maturing corn hybrid twice per year in remote or marginal areas. Furthermore, production may be extended up to three growing seasons in irrigated areas applying best agricultural practices. Another favorable ideotype for modern corn cultivation is short plant stature with low ear placement [63] to decrease the lodging percentage, to enable the increased planting density, to improve the light interception of the plant canopy, and to obtain higher economic yield [64]. The rapid adoption of those hybrid ideotypes mentioned above could help corn growers to minimize the risk of yield losses due to plant lodging during vegetative and grain-filling stages [65].

Eating quality is one of the major parameters in breeding specialty corns. Both taller kernel depth and good overall eating quality of selected hybrid 101L/TSC-10 × KV/mon offer satisfaction for consumers when enjoying steamed fresh corn. The concept of synergistic corn carrying multiple recessive genes was adopted in this study. As seen in Table 1, the sweet corn lines differed in mutant combinations from two (bt, wx) to three (bt, sh2, wx). The crosses between 101LBW (bt, wx) with 8 waxy corn lines (wx) produced 8 F1 hybrids with 3:1 expected segregation ratio of waxy and supersweet kernels, respectively, while the crosses between either 101L/TSC-4 or 101L/TSC-10 (bt, sh2, wx) with 8 waxy corn lines made 16 F1 hybrids with 9:7 segregation ratio. Sensory test on overall liking was included as selection criteria in this study. Excluding KV/mon x 101L/TSC-4, the top five crosses derived from the ORSI method came from three different waxy corn lines with the same sweet corn line 101L/TSC-10 carrying triple recessive genes. A previous study reported that combinations of three recessive genes had higher total sugar, sucrose, glucose, and fructose than those of two recessive genes [15]. However, the consumer preferences on modern waxy corn in Thailand are unique, as affected by the quality factors and eating cultures [66]. For example, people living in northeast Thailand consume sticky rice as a staple food and steamed waxy corn as an alternative meal. Therefore, efforts to improve the flavor like sweetness while maintaining the level of waxy taste or stickiness are recommended. Besides, ear appearances of husked F1 ears such as pericarp color and row arrangement are important. In Thailand, several commercial sweet-waxy corn hybrids differing in kernel colors have their niche markets, such as Khao Gam Wan (KGW) cultivar with dark purple kernels and Niyw Mwang Taem (NMT) cultivar with predominant white kernels and few violet kernels. Future research is suggested to include visual scoring of ear appearances of sweet-waxy corn hybrids harvested at the fresh stage to assess consumer perceptions.

5. Conclusions

Unweighted selection index and overall rank-sum index (ORSI) obtained preferable genetic gain on all observed traits ranging from low on row number per ear, anthesis and silking dates, kernel depth, and eating qualities to high on anthesis-silking interval and plant height difference. Both approaches were effective in simultaneous selection; however, the unweighted selection index became more efficient than ORSI, especially when there were many traits to evaluate and numerous genotypes to rank. The top five crosses derived from simultaneous selection performed good ideotypes such as high flowering synchrony, low ear position, moderate seed yield, taller male plant, early maturity, shorter hybrid plant stature, higher husked ear yield, and good eating quality, appealing to consumers, growers, and seed producers. Among selected crosses, genotype 101LTSC10 × KV/mon was promoted to be the promising candidate as a synergistic sweet-waxy corn hybrid cultivar.