George H. Shull coined the term heterosis in 1914 to

Transcription

1 Published online June 21, 2018 Agronomic Application of Genetic Resources Improving Breeding Efficiency of a Hybrid Maize Breeding Program Using a Three Heterotic-Group Classification XingMing Fan,* Yaqi Bi, Yudong Zhang, Daniel Jeffers, XingFu Yin, and Manjit Kang ABSTRACT Heterotic group classification has a crucial impact on maize (Zea mays L.) breeding efficiency in a hybrid-based breeding program. The objectives of this study were to: (i) investigate whether or not breeding efficiency could be improved using a three heteroticgroup (TriHG) classification (Reid and non-reid, and Suwan1 heterotic groups) system instead of a usual two heterotic-group (DiHG) classification (Reid and non-reid heterotic groups) system; and (ii) estimate the impact of TriHG and DiHG systems on breeding efficiency, utilizing the same data in computing specific breeding efficiency (SBE) and general breeding efficiency (GBE) in a subtropical breeding program in southern China. Twenty-five adapted tropical and subtropical lines were crossed to six testers using a line tester mating design. Data on grain yield were used to calculate SBE and GBE for comparing DiHGand TriHG-based strategies. The TriHG classification system increased GBE by 77.8% over the DiHG system without a significant loss in SBE. We concluded that the TriHG system was better than the DiHG system for improving maize-breeding efficiency. Therefore, maize breeders may re-think their breeding strategy for improving breeding efficiency in long-term breeding programs by adopting the suggested TriHG system. Hybrid maize improvement programs, especially in the early stages of breeding for identifying hybrids with high yield potential, can be become more efficient by using three testers with one from each of Reid, nonreid, and Suwan1 heterotic groups. This would improve the chances of developing high-yielding hybrids in a breeding program with the available, diverse germplasm. Core Ideas Maize breeding efficiency can be greatly improved by adopting three heterotic groups in hybrid breeding. Using three heterotic groups improves maize breeding efficiency for using genetic resources. General breeding efficiency is used to measure efficiency of genetic resource usage, and specific breeding efficiency is to obtain maximum output with inputs. Suwan1 is an important heterotic group from tropical/subtropical and should be explore more worldwide. Published in Agron. J. 110: (2018) doi: /agronj Supplemental material available online Available freely online through the author-supported open access option Copyright 2018 by the American Society of Agronomy 5585 Guilford Road, Madison, WI USA This is an open access article distributed under the CC BY-NC-ND license ( George H. Shull coined the term heterosis in 1914 to replace heterozygosis (Shull, 1948). Heterosis in maize (Zea mays L.) has been known for more than 100 yr and successfully utilized by the US hybrid maize industry for more than 80 yr now (Hallauer and Miranda, 1988; Duvick, 2001). The classification of germplasm into heterotic groups is important for developing high-yielding hybrids and improving breeding efficiency (Hallauer and Miranda, 1988; Barata and Carena, 2006; Fan et al., 2014). Melchinger and Gumber (1998) noted that for the exploitation of grain yield-heterosis in breeding, knowledge and understanding of both heterotic grouping and heterotic patterns were needed. They clearly distinguished between heterotic group and heterotic pattern as follows (Melchinger and Gumber, 1998): Heterotic group refers to a group of related or unrelated genotypes from the same or different populations, which display similar combining ability and heterotic response when crossed with genotypes from other genetically distinct germplasm groups. Heterotic pattern refers to a specific pair of heterotic groups that expresses high heterosis and high hybrid performance on crossing. For example, in the United States, predominantly two heterotic groups, classified as Reid derived from Iowa Stiff Stalk Synthetic (BSSS) and Lancaster (or non-reid), have been used. Crosses between these two heterotic groups have been the most exploited heterotic pattern (Hallauer and Miranda, 1988; Holley and Goodman, 1988; Hallauer and Carena, 2014); the Iodent represents the third heterotic group used by the maize seed industry (Nelson et al., 2008; Van Heerwaarden et al., 2012; Romay et al., 2013). Based on molecular-marker data, Romay et al. (2013) reported that Reid, non-reid, and tropical germplasm had distinct population structure, implying that introduced tropical maize germplasm might need to be X. Fan, Y. Bi, Y. Zhang, D. Jeffers, X. Yin, Yunnan Academy of Agricultural Sciences, Institute of Food Crops Sciences, 2238 Beijing Road, Kunming, Yunnan , China; M. Kang, Dep. of Plant Pathology, Kansas State Univ., Manhattan, KS Received 25 May Accepted 24 Jan *Corresponding author (xingmingfan@163. com). Abbreviations: CIMMYT, The International Maize and Wheat Improvement Center; DiHG, two heterotic groups; GBE, general breeding efficiency; GCA, general combining ability; GEM, germplasm enhancement of maize; GY, grain yield; IITA, International Institute of Tropical Agriculture; RCBD, randomized complete block design; SBE, specific breeding efficiency; SCA, specific combining ability; SNP, single nucleotide polymorphism; TriHG, three heterotic groups. Agronomy Journal Volume 110, Issue

2 classified into a distinct heterotic group that is different from the Reid, non-reid, and Iodent for more efficient use in hybrid breeding in the United States. The ability to efficiently determine the heterotic grouping and heterotic patterns as well as to identify new alternative heterotic groups for introduced germplasm is critical to the success of maize hybrid breeding programs (Richard et al., 2016). In Africa, historically, the most successful heterotic pattern has been based on two heterotic groups for white maize, i.e., Salisbury White (N) and Southern Cross (SC) (Olver, 1988). Crosses between inbred lines belonging to these two heterotic groups produced the single-cross hybrid SR52 in Zimbabwe (southern Rhodesia), which represented the first single-cross maize hybrid to be commercialized in Zimbabwe and formed the basis of maize breeding in Zimbabwe and eastern and southern Africa regions (Derera and Musimwa, 2015). The introduction and development of additional commercial white and yellow hybrids from temperate and tropical germplasm through regional collaborative programs operated by CIMMYT (The International Maize and Wheat Improvement Center) and IITA (International Institute of Tropical Agriculture) provided additional genetic diversity (Fan et al., 2003, 2010, 2016; Derera and Musimwa, 2015). Richard et al. (2016) suggested that to maximize the exploitation of heterosis, molecular markers can be used for classifying maize germplasm into heterotic groups and that the inclusion of temperate inbred lines derived from well-known heterotic groups could help in identifying potential heterotic groups in germplasm that have not yet been characterized. They demonstrated that the inclusion of inbred lines from established heterotic groups in a molecular characterization program was essential for proper identification of potential heterotic groups for Southern African maize inbred lines, and postulated that BSSS- and Lancasterderived lines could be used to increase divergence between N and SC groups; thus increasing the heterotic response. In a study utilizing single nucleotide polymorphism (SNP) markers, Wen et al. (2012) found that the genetic distance among tropical germplasm was much less than that between U.S. temperate and GEM (germplasm enhancement of maize) lines containing temperate and tropical germplasm, and indicated that, by introducing unique temperate alleles and distinct heterotic patterns, heterosis in tropical germplasm could be increased. In China, maize has been classified into two (Shi, 2007), three (Wang et al., 2011), four (Wu et al., 2007), or even more than four heterotic groups (Li et al., 2002, 2003). Lines derived from Suwan1, a widely utilized tropical population from Thailand, showed good combining ability for grain yield, but it could not be classified as either Reid or non-reid heterotic group (Fan et al., 2009). Thus, Fan et al. (2014) designated Suwan1 as a new heterotic group and proposed that maize lines should be classified into three heterotic groups (TriHG), i.e., Reid, non-reid, and Suwan1, and that this system would effectively exploit heterotic patterns for generating adapted hybrids, especially for use in southwestern China. Richard et al. (2016) also pointed out that classification of maize inbreds into known heterotic groups was one of the methods to reduce the number of duplicates while maintaining diversity and reducing the chances of evaluating a large number of undesirable crosses. Barata and Carena (2006) reported that proper heterotic grouping helped maximize combining ability and ultimately resulted in improved breeding efficiency. Breeding efficiency has been defined on the basis of several criteria, including the number of varieties released, selection gain per cycle, adoption indicators, and cost benefit analyses (Wang, 1997; Goshu, 2005; Ceccarelli, 2015). Ceccarelli (2015) used three measurements to compare the breeding efficiency between participatory breeding and conventional plant breeding: (i) the ratio of the number of varieties adopted (or released) to the number of crosses made; (ii) the response to selection; and (iii) cost benefit ratio. Depending on the number of heterotic groups being utilized in a breeding program, the first and third criteria are significantly affected by increased number of crosses, but if the heterotic grouping improves the identification of viable commercial hybrids, the per-hybrid cost will actually be reduced (Ceccarelli, 2015). Fan et al. (2014) showed that, based on GY data from 12 parental diallel data, addition of Suwan1 heterotic group to the two heterotic groups (DiHG) system improved maize breeding efficiency. However, the results from mathematical computations predicted that the DiHG should yield higher breeding efficiency than TriHG utilizing the same germplasm, which contradicted the results obtained in actual breeding (Fan et al., 2015). The data used for mathematical computation came from a diallel mating design involving 12 parents, where a parental line was repeatedly used. Because many of the high-yielding crosses were not independent, it was anticipated that the computed breeding efficiency would be biased. Further investigation in the modeling of breeding efficiency with a broader set of germplasm via a factorial design could help in determining if TriHG could be used for improving breeding efficiency. Therefore, the objectives of this study were: (i) to investigate whether or not use of the TriHG classification system would improve breeding efficiency over the DiHG classification system; and (ii) to contrast the effects of TriHG and DiHG systems relative to specific breeding efficiency (SBE) and general breeding efficiency (GBE) using a line tester mating design (mathematical definitions of SBE and GBE are given in the Materials and Methods section). Materials and Methods Germplasm Twenty-five tropical and subtropical maize inbred lines, possessing desirable agronomic traits, foliar disease resistance, and good adaptation to climatic conditions of Yunnan (China), were selected from germplasm introduced from CIMMYT, Mexico. To determine the value of the lines for use in hybrid breeding for southwestern China, the 25 lines were crossed in the summer of 2013 to six testers, two each from Suwan1, Reid, and non-reid heterotic groups (Table 1). The two testers representing each heterotic group were selected on the basis of the results obtained in previous experiments (Fan et al., 2000, 2002, 2008, 2014, 2016). Using a line tester mating design, 150 testcrosses were generated in Yunnan. Six plants from each of the 25 inbreds ( ) were pollinated with each tester ( ). Seed from the six ears was bulked and used for subsequent testcross evaluations. The 150 testcrosses and a commercial check, Yunrui 999, were evaluated in field trials during the summer of 2015 at Kunming (25 02 N, E; 1960 m above sea level), Wenshan (23 19 N, E; 1540 m above sea level), and Dehong (24 26 N, E; m above sea level). A 1210 Agronomy Journal Volume 110, Issue

3 Table 1. Population source (breeding start) and ecological adaptation of the 31 inbred lines used in the study. Lines Population source Ecological type Line s heterotic group L1 (P147-F2-102-S6/P33-C3-64-S4)-F2-B Mixed tropical/subtropical NA L2 (CML329/CML20)-F B-4 Mixed subtropical/tropical NA L3 (CML226/(CATETO DC1276/7619))B-B Mixed tropical/subtropical NA L4 (CML226/(CATETO DC1276/7619))B-B Mixed tropical/subtropical NA L5 (CML226/(CATETO DC1276/7619))B-B Mixed tropical/subtropical NA L6 (CML226/(CATETO DC1276/7619))B-B Mixed tropical/subtropical NA L7 (CML226/(CATETO DC1276/7619))B-B Mixed tropical/subtropical NA L8 (CML226/(CATETO DC1276/7619))B-B Mixed tropical/subtropical NA L9 (CML226/CATETO//CML323/CATETO)-F2-B-2-1-B Mixed tropical/subtropical NA L10 (P147-F2-152-S7/P45-C8-76-S9)-F2-B Mixed tropical/subtropical NA L11 GLSIY01HGA-B B Subtropical NA L12 FS8BT-278-B B/MD37)-8-B-32-1-B*3-3-B Mixed Temperate/subtropical NA L13 CLA155 = SA3-C4-FS(16/25) Tropical NA L14 CLA161 = SA4-C2-FS(21/26) Tropical NA L15 CL-RCY023 = (CL-02439/CML-286)-B B*8 Tropical NA L16 GLSIY01HGB B Subtropical NA L17 CML323/(CATETO DC1276/7619)-F B Mixed subtropical/tropical NA L18 (CML226/CATETO)-F2-B-1-2-B Mixed tropical/subtropical NA L19 P45-C6-FS B B Subtropical NA L20 P147-F2-114-S7/P45-C8-76-S9)-F2-B Mixed tropical/subtropical NA L21 (P147-F2-136-S7/P45-C8-76-S9)B-B-B Mixed tropical/subtropical NA L22 CLA44 = SA3-C4-FS(16/25) B Tropical NA L23 (P147-F2-136-S7/P45-C8-76-S9)B-B-B Mixed tropical/subtropical NA L24 P45-C B Subtropical NA L25 DTPY-C9-F B Tropical NA T1 Temperate Reid T2 Temperate Reid T3 Temperate non-reid T4 Temperate non-reid T5 Tropical Suwan1 T6 Tropical Suwan1 NA, not applicable. No information available about line s heterotic group. randomized complete-block design with three replications was used. Although other designs such as α lattice provides statistical efficiency, we used randomized complete block design (RCBD) for its simplicity as the field was uniform in fertility, which was confirmed by soil tests. Each plot consisted of a single 4-meter long row. Inter-row spacing was 0.7 m, whereas intra-row spacing was 0.25 m, resulting in a population density of ~55,300 plants ha 1. Compound fertilizer (N P 2 O 5 K 2 O, ) was applied at the rate of 1050 kg ha 1, a standard fertilizer rate for high-yield environments in the Yunnan province. Sulfur was applied at planting at the rate of 17 kg ha 1. Plots were irrigated once a week during the dry season; no irrigation was applied or needed during the rainy season. Plots were harvested at maturity, which was determined by black-layer formation. At maturity, a 10-ear sample was harvested from 10 consecutive plants from the middle of each row. Then grain yield per plant and other important agronomic traits were determined; yield was adjusted to a moisture content of 13% (130 g kg 1 ). Statistical Model and Data Analysis Combining ability analysis was conducted according to the model and method used by Fan et al. (2009). Data analyses were conducted by use of SAS software package (SAS Institute, 2005). Mean grain yield (GY) and specific combining ability (SCA) of all crosses were first computed from three replications for each location. Then the number of crosses that exceeded GY of the check by at least 5% was recorded, and this information was used for comparing breeding efficiency for the DiHG and TriHG classification systems. Mean GY and SCA were also compared between DiHG and TriHG systems. Paired t test was used to evaluate the differences in breeding efficiencies between DiHG and TRiHG. Tester Grouping For comparing breeding efficiency, the number of hybrids that exceeded the GY of the check by at least 5% in interheterotic (Reid non-reid; Reid Suwan1; and non-reid Suwan1) and intra-heterotic group (Reid Reid; non-reid non-reid; and Suwan1 Suwan1) crosses were recorded. For the TriHG system, there were three heterotic groups (i.e., Reid, non-reid, and Suwan1) and two testers in each heterotic group, designated as T1 and T2 for Reid, T3 and T4 for non-reid, and T5 and T6 for Suwan1 (Table 2). When three testers (TriHG), with one tester from each of the three heterotic groups, were used, eight combinations of testers resulted (shown as tester combinations 1 to 8 in Table 2). For the DiHG system, there were two heterotic groups (i.e., Reid and non-reid) and two Agronomy Journal Volume 110, Issue

4 Table 2. Tester group classification for possible combinations of testers from three heterotic group (TriHG) and two heterotic group (DiHG), with two testers per heterotic group. Testers are designated as T1 and T2 (Reid), T3 and T4 (non-reid, and T5 and T6 (Suwan1). Classification Tester Reid non-reid Suwan1 system combination T1 T2 T3 T4 T5 T6 TriHG 1 x x x 2 x x x 3 x x x 4 x x x 5 x x x 6 x x x 7 x x x 8 x x x DiHG 9 x x NA NA 10 x x NA NA 11 x x NA NA 12 x x NA NA DiHG, two heterotic group system; and TriHG, three heterotic group system. Tester combinations 1 to 12 are defined as a group of selected testers from three maize heterotic groups, i.e., Reid, non-reid, and Suwan1. We provide one example each for the TriHG and DiHG. TriHG example is represented by tester combination 1, where T1, T3, and T5 are used as testers; and DiHG example is represented by tester combination 9, where T1 and T3 are used as testers. NA, not applicable. testers in each heterotic group. In contrast, when two testers, with one tester from each of the two heterotic groups, were used, four combinations of testers were produced (shown as tester combinations 9 12) (Table 2). Heterotic Classification of the 25 Lines Used in the Study When different tester groups were utilized, the 25 lines were classified into different heterotic groups on the basis of specific combining ability and GY of a cross (SCA_GY). A line was assigned to the same heterotic group as the tester if the SCA of the cross was significant and negative or if GY was the lowest in the crosses between the line and the selected testers when the SCA was not statistically significant and negative. In total, 12 tester groups were used (Table 2); the heterotic groups for the 25 lines are presented in Supplementary Table S1 for tester group 1 and tester group 9 (as examples). Breeding Efficiency The numbers of crosses exceeding the GY of the check by at least 5% in inter-heterotic crosses and intra-heterotic crosses were recorded for all 12 tester combinations. Two measurements, GBE and SBE, which were used for comparing breeding efficiencies, were computed as follows: GBE (%) = (no. of hybrids that exceeded the GY of the check by 5% among inter-heterotic crosses in each of the 12 tester combinations 100)/no. of hybrids that exceeded the GY of the check by 5% among all crosses (i.e., 25 lines 6 testers). For the DiHG system, specific breeding efficiency was computed as follows: SBE (%) = (no. of hybrids that exceeded the GY of the check by 5% among inter-heterotic crosses 100)/total no. of inter-heterotic crosses made from the 25 lines 2 selected testers. For the TriHG system, specific breeding efficiency was computed as follows: SBE (%) = (no. of hybrids that exceeded the GY of the check by 5% among inter-heterotic crosses 100)/total no. of inter-heterotic crosses made from the 25 lines 3 selected testers. Specific combining ability and mean GY of the crosses between the selected testers from each of the 12 tester combinations and the 25 lines were computed to classify the 25 lines into either two heterotic groups (DiHG system) or three heterotic groups (TriHG system) using the SCA and GY (SCA_ GY). Examples of the computation and classification for the two tester groups are given in Supplemental Table S1 by using SCA_GY method as in other studies (Barata and Carena, 2006; Fan et al., 2008, 2015). Results Differences in General Breeding Efficiency and Specific Breeding Efficiency between Two Heterotic Group and Three Heterotic Group Systems The numbers of high-yielding crosses recorded for inter-heterotic and intra-heterotic groups and values for GBE and SBE are shown in Table 3. Tester combinations 1 and 9 are used below as examples to show the computations for GBE and SBE. Twenty-three out of the 41 high-yield crosses (data not shown) among the 150 total crosses belonged to tester combination 1 (TriHG) (Table 3). Therefore, GBE = 23/41 = For three testers, with one tester taken from each of the three heterotic groups, a total of 50 inter-heterotic crosses resulted. Therefore, SBE = 23/50 = Eleven out of a total of 41 high-yield crosses belonged to tester group 9 (DiHG). Therefore, GBE = 11/41 = The SBE for this tester combination = 11/25 = 0.44 When all possible tester combinations were included for TriHG (with one tester from each heterotic group), the GBE ranged from 0.29 to 0.63, with a mean of For the DiHG (with one tester from each heterotic group), GBE ranged from 0.24 to 0.32, with a mean of In contrast, the mean SBE for the TriHG was and mean SBE for the DiHG was The same trend was observed when the yield data were analyzed by location (Supplementary Table S2). The results in Table 3 revealed the following: 1. On the average, the SBE with the TriHG system was about 10% less than that with the DiHG system [( ) 100/0.44 = 10.2%]. These results were consistent with the theoretical deduction (Fan et al., 2014), in that the larger the number of heterotic groups maize lines are classified into, the lower the SBE. The paired t test showed that the difference in SBE between the DiHG and TriHG, however, 1212 Agronomy Journal Volume 110, Issue

5 Table 3. Breeding efficiencies for possible combinations of testers of two heterotic group (DiHG) and three heterotic group (TriHG) classification systems, based on numbers of crosses with grain yield (GY) 5% higher than GY of check (or 164 g plant 1 ). System Tester combination Inter-group Intra-group Total General breeding efficiency Specific breeding efficiency TriHG Mean DiHG Mean DiHG, classification according to two heterotic groups; and TriHG, classification according to three eterotic groups. Tester combinations 1 to 12 are defined as a group of selected testers from Reid, non-reid, and Suwan1 heterotic groups. The probabilities of the mean comparison between TriHG and DiHG for general breeding efficiency (GBE) and specific breeding efficiency (SBE) are and 0. 35, respectively. The GBE for tester combination 1 = 23/41 = (41 is the number of total hybrids with 5% higher grain yield than the check). The SBE = 23/50 = 0.46 (50 is the number of total inter-heterotic crosses made between the 3 testers and 25 lines with known heterotic groups as defined in Supplemental Table S1. was not statistically significant at the α = 0.05 level. 2. The GBE for the TriHG system was 77.8% [( ) 100/0.27 = 77.8%] higher than that for the DiHG system, which was highly significant (P < 0.01) according to the paired t test. 3. As the crosses locations interaction was significant (ANOVA not shown), we computed GBE and SBE by location (Supplementary Table S2), which gave results similar to those in Table 3. Mean Grain Yield and Specific Combining Ability of High-Yield Crosses for Two Heterotic Groups and Three Heterotic Groups The results for mean SCA effects in both inter-heterotic and intra-heterotic crosses obtained for the DiHG and TriHG systems are presented in Table 4. These results revealed the following information: 1. No significant differences existed between the TriHG and DiHG for mean GY and mean SCA. 2. Though mean SCA effects were statistically different (P < 0.01) between inter-heterotic and intra-heterotic crosses, no significant differences were detected between inter-heterotic crosses and intra-heterotic crosses for GY in the high-gy group (Table 4). Grain Yield of Crosses with Lines from Suwan1 as One of the Parents in Crosses The GY values for Reid non-reid, Reid Suwan1, and non-reid Suwan1, and their reciprocal crosses, are plotted in Fig. 1. The results showed that Reid non-reid crosses had the lowest mean GY, which was significantly lower (P < 0.01) than those of non-reid Suwan1, Suwan1 Reid, and Suwan1 non-reid crosses. The non-reid Suwan1 and Suwan1 Reid crosses had the highest GY. These results revealed that the Suwan1 heterotic group was responsible for the higher breeding efficiency of the TriHG compared with the DiHG system. In addition, non-reid Reid and Suwan1 Reid crosses had significantly higher GY than their reciprocal crosses (P < 0.01). Discussion Three Heterotic Groups Improved Breeding Efficiency over Two Heterotic Groups In this study, breeding efficiency was examined using two measurements, GBE and SBE. Our results showed that there were significant differences between the breeding efficiencies of DiHG and TriHG systems (Table 3, Supplementary Table S2). The results clearly demonstrated the following: 1. The TriHG system had significantly higher GBE than the DiHG system. As GBE measures the efficiency of utilization of available germplasm, the TriHG system would be preferred over the DiHG system. 2. The SBE did not significantly differ for the TriHG and the DiHG systems, indicating that there was no significant difference between these two systems in the utilization of resources, such as labor, time, capital, etc. Fan et al. (2014) demonstrated that, theoretically, the larger the number of heterotic groups used, the smaller the SBE. From the present study, the overall conclusion would be that TriHG should be preferred to DiHG for improving the breeding efficiency in utilization of genetic resources. Results from Fig. 1 suggested that the addition of Suwan1 heterotic group, i.e., three heterotic groups (TriHG) instead of two heterotic groups (DiHG) in a maize breeding program, could substantially increase the chances of identifying new hybrids with GY higher than that of the check. This should lead to an improvement in breeding efficiency of the maize improvement program in the subtropical region where the evaluations were performed. Adaptation of tropical germplasm and maturity could be responsible for a significant portion of the observed improvement relative to GBE. This study, with Agronomy Journal Volume 110, Issue

6 Table 4. Mean grain yield (Avg_GY) and mean specific combining ability (Avg_SCA) for inter-heterotic crosses (inter-crosses) and intraheterotic crosses (intra-crosses between top group and bottom group based on grain yield, GY). Top group (GY > 5% check) Low group (GY 5% of check) Inter-cross Intra-cross Inter-cross Intra-cross Tester combination Avg_GY Avg_SCA Avg_GY Avg_SCA Avg_GY Avg_SCA Avg_GY Avg_SCA NA NA NA NA NA NA Avg_TriHG Avg_DiHG P for mean comparison Tester combinations 1 to 12 are defined as a group of selected testers from Reid, non-reid, and Suwan1 heterotic groups. Avg_TriHG, mean across the eight tester combinations in the table; and Avg_DiHG, mean across the four tester combinations in the table. NA, no cross with GY > 5% check. testers from the three heterotic groups, produced testcross hybrids with a range of maturities and adaptive traits, including reaction to foliar diseases, such as turcicum leaf blight [caused by Exserohilum turcicum (Pass) K.J. Leonard] and gray leaf spot [incited by Cercospora zeina Crous & U. Braun.], and Gibberella ear rot [incited by Gibberella zeae (Schwein) Petch; also known Fusarium graminearum (Schwabe)]. Utilization of the Suwan1 testers resulted in parental inbreds with less susceptibility to foliar diseases and Gibberella ear rot in test crosses than that of the Reid and non-reid testers (Sriwatanapongse et al., 1993). When a cross involved only tropical and subtropical germplasm, hybrid maturity was delayed, which lengthened the growing season and provided increased opportunity for achieving improved yields. The subtropical region of southwestern China has a relatively long growing season, and hybrids containing a parent from the Suwan1 heterotic group could result in improved performance, but depending on the elevation, they might not be desirable when both parents are of non-temperate origin because of late-maturity issues. Results also showed (Fig. 1) that when the male parent belonged to the Reid heterotic group (e.g., non-reid Reid and Suwan1 Reid), resulting crosses had significantly higher GY than their reciprocal crosses (P < 0.01). These results suggested that it would be a good idea to pay attention to the direction of crosses in breeding programs. Reciprocal effects may mean the existence of favorable interactions between cytoplasmic factors in lines from Suwan1 and non-reid heterotic groups, and the nuclear genes from Reid lines. From a 12-parent diallel analysis, Fan et al. (2014) found that reciprocal crosses could significantly impact GY, SCA, and GCA estimates, and that the reciprocal effects were highly correlated with non-maternal effects, which represented interactions between nuclear genes and cytoplasmic factors (Zhang and Kang, 1997; Zhang et al., 2005; Mahgoub, 2011). The GY and SCA estimates were found to be most affected by reciprocal effects. Fan et al. (2014) suggested that additional studies would be needed to examine the effects of cytoplasmic genes on GY and other agronomic traits. Conclusions The current study examined the effects of TriHG system vs. DiHG system in the context of improving efficiency of a breeding program. The TriHG system improved GBE by 78% over the DiHG system. This provided the possibility of identifying additional top-performing testcrosses with the same inbred lines used in a breeding program. Some potential applications of the results of the present study are as follows: 1. As the TriHG system improved GBE over the DiHG system, three testers (with one tester each from the three established maize heterotic groups) should be used in hybrid maize improvement programs, especially in the early stages of breeding for identifying hybrids with high yield potential. This should improve the chances of developing high-yielding hybrids in a breeding program with the available, diverse germplasm. This recommendation should be especially useful for non-temperate maize breeders trying to improve breeding efficiency in developing yellow maize hybrids. With the Suwan1 heterotic group included in breeding efforts for non-temperate areas germplasm, especially CIMMYT-derived lines can be classified more efficiently since many do not group well with Reid and non-reid. Recent breeding activities are exploiting both temperate patent expired germplasm from the United States as well as CIMMYT-improved lines in line and hybrid development for non-temperate environments. Both yellow and white segregants are being utilized in the breeding programs. However, the same concept could be applied to a white maize breeding program for Africa or Latin America, where both temperate and non-temperate maize germplasm are utilized. 2. The Suwan1-derived lines have been shown to possess good 1214 Agronomy Journal Volume 110, Issue

7 Fig. 1. Mean grain yield of non-reid Suwan1 (NS), Suwan1 Reid (SR), non-reid Reid (NR), Suwan1 non-reid (SN), Reid Suwan1 (RS), and Reid non-reid (RN) crosses. disease resistance, lodging resistance, drought tolerance, and high GY potential, but they are usually not desirable for improving plant and ear height, and grain dry-down rate, and have poor adaptation to temperate environments because of photoperiod sensitivity (Fan et al., 2003, 2009, 2015). In contrast, the inbred lines derived from Reid and non-reid heterotic groups are desirable for reducing plant and ear height, enhancing grain dry-down rate, and improving adaptation to temperate environments, but often lack sufficient resistance/tolerance to biotic and abiotic stresses, and lodging resistance. Reid and non-reid germplasm were developed from non-temperate maize introduced from Mexico, the Caribbean, and Central and South America (Hallauer and Miranda, 1988; Holley and Goodman, 1988; Goodman, 2005). The selection for adaptation and subsequent hybrid maize breeding programs have narrowed the genetic base of the commercialized Reid and non-reid germplasm (Goodman, 2005; Van Heerwaarden et al., 2012, Hallauer and Carena, 2014). Thus, frequently, there is lack of genetic diversity for stable resistance to biotic stresses. A backcross breeding program would be an effective way to transfer useful genes from temperate heterotic groups (Reid and non-reid) into tropical germplasm, such as Suwan1, and vice versa (Nelson and Goodman, 2008; Chen et al., 2010). The introgression of genes from exotic germplasm via the backcross procedure usually does not change the recurrent parent s original heterotic group (Fan et al., 2015). For effectively utilizing germplasm from the Suwan1 heterotic group to broaden the genetic base of the local maize germplasm, introgression of specific genes from local lines into Suwan1 may be needed to improve some of its undesirable attributes (Fan et al., 2000, 2002, 2014; Chen et al., 2005, 2010). Improved lines from Suwan1 heterotic group can be used directly for developing maize hybrids when maturity is not a limitation, or for improving local Reid and non-reid germplasm for biotic stress resistance (Xu et al., 2014; Liu et al., 2016), lodging resistance, and high GY potential. Based on TriHG system, a breeding procedure for improving Suwan1 germplasm could be as follows: Assuming the A B cross has high GY, where line A is from Reid and line B from non-reid heterotic group and both lines lack resistance to a particular disease that prevents it from being released as a commercial hybrid, we could take line C developed from Suwan1 that possesses a high level of resistance to the disease, and cross it with either line A or line B to transfer the disease resistance genes. Because the heterotic group of a line following backcrossing (to at least BC3 level) is usually not changed (Chen et al., 2010; Fan et al., 2015), the GY heterosis of A B will likely not change with the introgression of disease resistance genes into either line A or B, and GY of the cross would be expected to provide improved yield stability in the presence of the disease. This should greatly increase the chances of high-yield hybrids being developed, released, and commercialized. 3. The results from the present study strengthened the validity of utilizing the TriHG system in maize breeding programs in China and possibly elsewhere. Data from a long-term selection study (personal communication, K.R. Lamkey et al., 2004) involving Iowa Stiff Stalk Synthetic (BSSS) and Iowa Corn Borer Synthetic no. 1 (BSCB1), revealed that selection could even create new heterotic groups. As the TriHG system was found to improve breeding efficiency over the DiHG system, and because selection could possibly change the heterotic groups of maize lines, the results of the present study should encourage maize breeders to re-think their breeding strategy for improving breeding efficiency in their long-term breeding programs by adopting the suggested TriHG system, where possible. Supplemental Material Table S1. The heterotic classification for the 25 lines based on grain yield (GY) and specific combining ability (SCA) with selected testers. Table S2. Breeding efficiencies for possible combinations of testers of two heterotic group (DiHG) and three heterotic group (TriHG) classifications based on numbers of crosses with grain yield (GY) 5% higher than GY of check (or 164 g plant 1 ) at three different locations. Acknowledgments We acknowledge the financial support from Yunnan Leading Talent Introduction Project (2014HA002), the Crucial New Production Project of Yunnan (2012BB012), and Major Research and Development Program of China (2016YFD ). We also acknowledge the support given by CIMMYT in supplying the lines included in this study. We thank Cargill International, which provided to CIMMYT the CATETO double-cross hybrid for use in their breeding program. We also thank personnel from experimental locations in Wenshan and Dehong for their field assistance. References Barata, C., and M. Carena Classification of North Dakota maize inbred lines into heterotic groups based on molecular and testcross data. Euphytica 151: doi: /s y Ceccarelli, S Efficiency of plant breeding. Crop Sci. 55: doi: /cropsci Chen, H.M., Y.D. Zhang, W. Chen, M.S. Kang, J. Tan, Y.F. Wang, J.Y. Yang, and X.M. Fan Improving grain yield and yield components via backcross procedure. Maydica 55: Chen, Z.H., X. Gao, and Y.F. Zhu Study on the combining ability and heterosis between Suwan and four major maize germplasm of China. (In Chinese, with English abstract). J. Maize Sci. 13:5 9. Derera, J., and T.R. Musimwa Why SR52 is such a great maize hybrid? I. Heterosis and generation mean analysis. Euphytica 205: doi: /s Duvick, D.N Biotechnology in the 1930s: The development of hybrid maize. Nat. Rev. Genet. 2: doi: / Agronomy Journal Volume 110, Issue

8 Fan, X.M., Y.Q. Bi, Y.D. Zhang, D.F. Jeffers, W.H. Yao, H.M. Chen, L.Q. Zhao, and M.S. Kang Use of the Suwan1 heterotic group in maize breeding programs in Southwestern China. Agron. J. 107: doi: /agronj Fan, X.M., H.M. Chen, J. Tan, C.X. Xu, Y.M. Zhang, Y.X. Huang, and M.S. Kang A new maize heterotic pattern between temperate and tropical germplasm. Agron. J. 100: doi: / agronj Fan, X.M., J. Tan, and J.Y. Yang Utilization of exotic tropical and subtropical maize germplasm. (In Chinese, with English abstract). Xi Nan Nong Ye Xue Bao 13: Fan, X.M., J. Tan, J.Y. Yang, F. Liu, B.H. Huang, and Y.X. Huang Study on yield combining ability and genetic relationship between exotic tropical, subtropical maize inbreeds and domestic temperate inbreeds. (In Chinese, with English abstract). Sci. Agric. Sin. 35: Fan, X.M., X.F. Yin, Y.D. Zhang, Y.Q. Bi, L. Liu, H.M. Chen, and M.S. Kang Combining ability estimation for grain yield of maize exotic germplasm using testers from three heterotic groups. Crop Sci. 56: doi: /cropsci Fan, X.M., S.H. Zhang, J. Tan, M.S. Li, and X.H. Li Heterotic grouping of quality protein maize inbreds by SSR markers. (In Chinese, with English abstract). Acta Agron. Sin. 29: Fan, X.M., Y.D. Zhang, L. Liu, H.M. Chen, W.H. Yao, M.S. Kang, and J.Y. Yang Screening tropical germplasm by temperate inbred testers. Maydica 55: Fan, X.M., Y.D. Zhang, W.H. Yao, Y.Q. Bi, L. Liu, H.M. Chen, and M.S. Kang Reciprocal diallel crosses impact combining ability, variance estimation, and heterotic group classification. Crop Sci. 54: doi: /cropsci Fan, X.M., Y.M. Zhang, W.H. Yao, H.M. Chen, J. Tan, C.X. Xu, X.L. Han, L.M. Luo, and M.S. Kang Classifying maize inbred lines into heterotic groups using a factorial mating design. Agron. J. 101: doi: /agronj Goodman, M.M Broadening the U.S. maize germplasm base. Maydica 50: Goshu, G Breeding efficiency, lifetime lactation and calving performance of Friesian-Boran crossbred cows at Cheffa farm, Ethiopia. Livest. Res. Rural Dev. 17:73 gosh17073.htm (accessed 7 Nov. 2017). Hallauer, A.R., and M.J. Carena Adaptation of tropical maize germplasm to temperate environments. Euphytica 196:1 11. doi: /s Hallauer, A.R., and J.B. Miranda Quantitative genetics in maize breeding. 2nd ed. Iowa State Univ. Press, Ames, IA. Holley, R.N., and M.M. Goodman Yield potential of tropical hybrid maize derivatives. Crop Sci. 28: doi: /cropsc i x x Li, M.S., S.H. Zhang, X.H. Li, G.T. Pan, L. Bai, and Z.B. Peng Study on heterotic groups among maize inbred lines based on SCA. (In Chinese, with English abstract). Sci. Agric. Sin. 35: Li, X.H., L.X. Yuan, Z.H. Li, S.H. Zhang, M.S. Li, and W.H. Li Heterotic grouping of 70 maize inbred lines by SSR markers. (In Chinese, with English abstract). Sci. Agric. Sin. 36: Liu, L., J. Tan, Y.D. Zhang, H.Y. Li, Y.Q. Bi, L.J. Yu, D.P. Jeffers, M.S. Kang, and X.M. Fan QTL mapping for gray leaf spot resistance in a tropical maize population. Plant Dis. 100: doi: /pdis re Mahgoub, G.M.A Partitioning of general and specific combining ability effects for estimating maternal and reciprocal effects. J. Agric. Sci. 3: Melchinger, A.E., and R.K. Gumber Overview of heterosis and heterotic groups in agronomic crops. In: K.R. Lamkey and J.E. Staub, editors, Concepts and breeding of heterosis in crop plants. CSSA, Madison, WI. p Nelson, P.T., N.D. Coles, J.B. Holland, D.M. Bubeck, S. Smith, and M.M. Goodman Molecular characterization of maize inbreds with expired U.S. plant variety protection. Crop Sci. 48: doi: /cropsci Nelson, P.T., and M.M. Goodman Evaluation of elite exotic maize inbreds for use in temperate breeding. Crop Sci. 48: doi: /cropsci Olver, R.C Maize breeding in Zimbabwe: In: H.O. Gevers. Proceedings of Eleventh South African Maize Breeding Symposium, p Richard, C., D.S. Osiru, M.S. Mwala, and T. Lubberstedt Genetic diversity and heterotic grouping of the core subset of southern African and temperate maize (Zea mays L.) inbred lines using SNP markers. Maydica 61:M3. p Romay, M.C., M.J. Millard, J.C. Glaubitz, J.A. Peiffer, K.L. Swarts, T.M. Casstevens, R.J. Elshire, C.B. Acharya, S.E. Mitchell, A.A. Flint- Garcia, M.D. McMullen, J.B. Holland, E.S. Buckler, and C.A.C. Gardner Comprehensive genotyping of the USA national maize inbred seed bank. Genome Biol. 14:R55. SAS Institute SAS/STAT user s guide. 2nd ed. SAS Institute, Cary, NC. Shi, L The impacts of US germplasm on maize breeding efforts in China. (In Chinese, with English abstract). J. Maize Sci. 15:1 4. Shull, G.H What is heterosis? Genetics 33(5): Sriwatanapongse, S., S. Jinahyin, and S.K. Vasal Suwan-1: Maize from Thailand to the world. CIMMYT, Mexico. Van Heerwaarden, J., M.B. Hufford, and J. Ross-Ibarra Historical genomics of North American maize. Proc. Natl. Acad. Sci. USA 109: Wang, G.S., J.L. Chen, W. Hou, B.L. Yan, H.B. Sun, and P.H. Li Study on heterotic group and heterotic pattern among maize inbred lines based on SCA. (In Chinese, with English abstract). Mod. Agric. Sci. Technol. 3: Wang, J More efficient breeding systems for controlling inbreeding and effective size in animal populations. Heredity (Edinb) 79: doi: /hdy Wen, W., T. Guo, V.H.C. Tovar, H. Li, J. Yan, and S. Taba The strategy and potential utilisation of temperate germplasm for tropical germplasm improvement: A case study of maize (Zea mays L.). Mol. Breed. 29: doi: /s Wu, J.C., C.X. Xu, H.M. Chen, J. Tan, X.R. Han, B.H. Huang, and X.M. Fan Studies on combining ability and heterotic grouping of 24 quality protein maize inbreds and four temperate representative inbreds of Chinese major heterotic groups. (In Chinese, with English abstract). Sci. Agric. Sin. 40: Xu, L., Y. Zhang, S.Q. Shao, W. Chen, J. Tan, M. Zhu, T. Zhong, X.M. Fan, and M.L. Xu High-resolution mapping and characterization of qrgls2, a major quantitative trait locus involved in maize resistance to gray leaf spot. BMC Plant Biol. 14:230. doi: / s Zhang, Y., and M.S. Kang DIALLEL-SAS: A SAS program for Griffing s diallel analyses. Agron. J. 89: doi: /agronj x Zhang, Y., M.S. Kang, and K.R. Lamkey DIALLEL-SAS05: A comprehensive program for Griffing s and Gardner-Eberhart analyses. Agron. J. 97: doi: /agronj Agronomy Journal Volume 110, Issue