Evaluating General Combining Ability for Multiple Traits in Tetraploid Bahiagrass

Florencia Marcón Northeast Institute of Botany: Instituto de Botanica del Nordeste Elsa Andrea Brugnoli (  abrugnoli@agr.unne.edu.ar ) Northeast Institute of Botany: Instituto de Botanica del Nordeste José A. Rodrigues Nunes Federal University of Lavras: Universidade Federal de Lavras Valeria A. Gutierrez Universidad Nacional del Nordeste Eric J. Martínez Northeast Institute of Botany: Instituto de Botanica del Nordeste Carlos A. Acuña Northeast Institute of Botany: Instituto de Botanica del Nordeste

The most important forage species used for beef cattle production systems in the tropics 2 and subtropics reproduce asexually by apomixis (Jank et al. 2014). Among them the genera 3 Panicum, Urochloa, Paspalum and Cenchrus ) have the greatest economic 4 value. Apomixis allows cloning superior genotypes by seed without loss of vigor or change in 5 genotype through generations (Hanna and Bashaw 1987), which provides the opportunity to 6 exploit heterosis. 7 The most popular breeding method in apomictic species has been direct selection from wild 8 accessions. However, finding new accessions that combine multiple traits of interest is a limitation 9 (Miles 2007). Hybridization arose as a new way to improve apomictic species once sexual plants 10 with the same ploidy level of apomictic were found in nature or were artificially created (Vogel 11 and Burson 2004). The main objective of this breeding approach is to release the natural diversity 12 contained in apomictic ecotypes and fix superior F1 hybrids (Miles 2007). 13 Paspalum notatum Flüggé, bahiagrass, is the primary constituent of rangelands of South 14 America and it is cultivated as forage and turf around the world (Gates et al. 2004). In the United 15 States, bahiagrass is the main pasture used in the beef cattle production systems of the Coastal 16 Plain region (Blount and Acuña 2009). This species has tetraploids that reproduce by apomixis 17 and diploids that reproduce sexually (Ortiz et al. 2013). The apomictic tetraploid genotype is the contrasting dominant allele frequencies for traits of interest is crucial to exploit hybrid vigor. 2 Considering the importance of heterosis in apomictic species and that heterotic parents are 3 required, Miles (2007) proposed recurrent selection based on combining ability to improve a 4 sexual synthetic tetraploid population of Brachiaria by the accumulation of heterotic effects over 5 selection cycles. And then obtain superior apomictic hybrids by crossing the improved sexual 6 genotypes with apomictic ones. Based on the scheme proposed by Miles (2007), Marcón et al. 7 (2020) carried out recurrent selection based on general combining ability to improve the SSTP of 8 bahiagrass. This breeding approach consisted in a four-step process where (1) plants of the SSTP 9 where crossed by a group of elite apomictic genotypes used as testers, (2) superior sexual parents 10 were selected on the basis of the performance of their testcross progeny, (3) the selected sexual 11 parents were polycrossed to form a recombinant superior population, and (4) sexual plants of the 12 recombined population were crossed by superior apomictic plants. 13 The testcross progeny test was successful in improving tetraploid bahiagrass for forage    1996). Loss of information could lead to a wrong estimation which might reduce the selection 21 efficiency. 22 In order to make a more efficient election of the best recombinant parents in recurrent 1 selection schemes based on progeny test of apomictic forages, Worthington and Miles (2015) 2 proposed to use best linear unbiased predictor (BLUP) to predict genetic effects such as general 3 and specific combining ability. BLUP is a predictor of random effects and it is part of the mixed 4 model approach that may be more informative than analysis of variance (Bernardo 2010). BLUP

18
Plant material 19 A group of 29 half-sib families of tetraploid P. notatum was used in this study. These   Mixed model approach analysis and narrow-sense heritability 9 Data from the half-sib family evaluations were analyzed using a mixed model approach

15
The parameters 2 , and 2 refer to the variance components associated with family, and error, 16 respectively. ⁄ , where 2 is the 5 genetic variance among half-sib families, which corresponds to ¼ of the additive genetic variance, 6 and ℎ 2 is the average phenotypic variance (Falconer and Mackay 1996). The analysis was carried 7 out with R software (R Core Team 2019).   The value of the selection index for each family was calculated as follow: where ̅ is the scaled mean family BLUP of the k-th trait, and is the weight associated with 10 the k-th trait, as aforementioned. Therefore, the selection index varied from 0 to 1, where 1 11 represented the best family for all evaluated traits and 0 represented the worst family.

13
Mixed model approach analysis and narrow-sense heritability 14 Narrow sense heritability (h 2 ) on family-mean basis varied from 0.56 to 0.86 (Fig 1). 15 Flowering index and plant height showed the greatest values of h 2 (0.86 and 0.83, respectively) 16 whereas the RT/VT ratio and winter regrowth exhibited the lowest values (0.56 and 0.65, 17 respectively). 18 The BLUPs of the families and their prediction confidence intervals showed that some of 19 those were statistically superior to the mean for all evaluated traits (Fig 2), except for summer 20 forage yield. For cold tolerance, families 21, 28, 5, 25 and 29 were statistically over the mean. and summer forage yield, just two families were over the mean value (Fig 2). Families 9, 5, 16 and statistically over the mean (Fig 2). 7 Biplot analysis and correlation test 8 Analyzing the biplot obtained using the mean BLUP values of each family, high variability 9 among families was observed and that all traits contributed almost equally to this variation, except 10 for seed yield (Fig 3). Family 9 and 16 stood out for their high summer growth and RT/VT and  (Fig 3). However, no multivariate pattern defines the best family for all the traits 14 simultaneously. Combining BLUP and biplot analysis, sexual parents of families 9, 5, 8, 21, 16 15 and 28 exhibited greater general combining ability. 16 In addition, most of the genetic correlations between the traits were positive (Fig 4). The and leaf blade length and RT/RV (0.73) (Fig 4). New breeding schemes such as recurrent selection based on general combining ability have 9 been successfully used in improving tetraploid bahiagrass. However, one of the challenges of these 10 methods is efficiency to select superior parents to form a new recombinant population. For this 11 reason, we employed the BLUP procedure in order to estimate accurately the general combining 12 ability of sexual parents based on their progeny performance.

13
BLUP is a predictor of random effects with desirable properties for plant breeding purposes 14 and it is part of the mixed model approach (Bernardo 2010). This analysis has been successfully 15 used in major crops, such as maize, to select superior parents based on the performance of their 16 progeny and continue with the breeding process (Bernardo 1999). In single crosses of Urochloa 17 humidicola, an apomictic forage species, this analysis allowed identifying the most promising also must establish quickly after sowing or planting and grow rapidly during the growing season 8 (Burton 1986). Thus, initial growth, spring forage yield and summer growth were among the most 9 important forage characteristics. Another important aspect in forage breeding is cultivar increase 10 or reseeding (Burton 1986) so seed yield played an important role. All these traits showed a large 11 variation among families (Fig 2) and also exhibited a medium to high narrow-sense heritability 12 (Fig 1) which means that it would be possible to improve tetraploid bahiagrass for those traits. yield. In addition, BLUP considers the genotypes as random effects while analysis of variance as 21 fixed so lot of information might have been missed and resulted in a less efficient selection.

22
A biplot analysis was also conducted to combine all traits in a single analysis and make a 1 more precise selection (Fig 3). Biplot did not show a clear pattern of the best families for all traits. that selection index based on mean BLUP might be use as a tool to estimate general and specific 15 combining ability in recurrent selection schemes. 16 Analyzing data generated by BLUP, biplot and selection index, it was observed that the 17 three methods allowed detecting the same superior female parents, which may be demonstrating 18 that these three analysis were successful to estimate general combining ability in tetraploid 19 bahiagrass. In addition, the joint analysis tool allowed to increase the efficiency of the selection of 20 the best parents which might be because the multivariate approaches taking account more properly bahiagrass. However, in this study the correlation values were higher. Furthermore, these traits 5 exhibited a high narrow-sense heritability (Fig 4), which means that plant height and leaf blade 6 length may be use to select indirectly plants that tolerate and growth during the winter season, the  In conclusion, BLUP, biplot and selection index were successful in estimating general 12 combining ability of a group of sexual parents of tetraploid bahiagrass based on the performance 13 of their progeny. The three methods were equally effective since they identified the same best 14 female parents that could be used to continue with the breeding program. The joint use of those 15 methods allowed identifying in a more efficient way sexual parents with greater general combining 16 ability, so in apomictic species a more accurate selection may be carried out by using more than 17 one analysis. High positive correlated traits with high narrow-sense heritability may be indicating 18 that it is possible to select some traits indirectly by recurrent selection methods such as recurrent 19 selection based on combining ability.    and principal components to select sexual genitors and hybrids of Urochloa decumbens. 22 Crop Breed Appl Biot 20:e28082022.