Combining capacity and heterosis in eggplant hybrids under high temperatures

The objective of this work was to estimate the combinatorial capacity and heterosis of eggplant hybrids under high temperature conditions. Seven genitors, twelve hybrid combinations, originated from a partial diallel, and the Ciça F1 hybrid, as control, were evaluated. The experiment was conducted under greenhouse conditions in randomized block design with four replications, from April to December 2017. The assessed traits related to high temperatures were pollen viability (PV) and fruit fixation index (FFI); the morphoagronomic traits were number of fruits per plant (NFP), fruit weight (FWe), production per plant (PP), fruit length (FL), fruit width (FWi), fruit length/width ratio (FLWR) and plant height (PH). The variance analysis showed greater participation of the additive gene effects in relation to the non-additive gene effects in most traits, except for PV. The genitors CNPH 141, CNPH 135, CNPH 109 and CNPH 51 stood out with favorable gene effects to obtain genotypes tolerant to high temperatures, since they present good general combining ability (GCA) for the traits FFI, NFP and PP. The 1x4 and 3x4 hybrids presented positive estimates for both GCA and specific combining ability (SCA), demonstrating a greater potential to be used in breeding to increase the FFI, NFP and PP, under high temperatures. The 1x4, 1x5 and 1x6 hybrids expressed positive heterosis for most analyzed traits. The 1x4 hybrid stood out for the highest averages for PV, FFI, NFP and PP. For FWe, FL, FWi and FLWR, both positive and negative heterosis were observed, as consequence of the phenotypic variability of the genitors for these traits and suggests the possibility of selection for different sizes and formats.

Eggplant is one of the most demanding vegetables for high temperature, with high sensitivity to cold and frost, during flowering and fruiting it tolerates milder temperatures (Polverente et al., 2005). The ideal temperature for crop growth and development is between 22 and 30°C, the decrease to 17°C results in inhibition of plant development (Adamczewska-Sowińska & Krygier, 2013). Flower abortion is favored by the natural reduction of daylight and high temperature at night (30°C) (Saito & Ito, 1973). Productivity is drastically reduced when temperature exceeds 32°C (Baswana et al., 2006).
In northeastern Brazil, especially when flowering coincides with the hottest period of the year, high temperatures have been a limiting factor for productivity, increasing the occurrence of malformation and/or fruit abortion, especially on crops under greenhouse conditions, where internal temperatures are higher than the outside one, causing a considerable reduction on crop yield in the region (Valadares et al., 2019). It is therefore important that high temperature tolerant eggplant strains and hybrids are obtained and evaluated under such conditions. Eggplant hybrids can be obtained from crossbreeding involving malesterile strains or by manual emasculation and pollination (George, 2009). In the crossing stage, it is of fundamental importance to obtain information about combinatorial capacity, the per se potential of the strains, and estimates of heterosis of the hybrid combinations obtained, in order to discriminate superior genitors for hybridization in a breeding program. To meet this purpose, diallel crosses are used, whose analyzes allow estimating the general and specific combining ability (Griffing, 1956), making inferences about heterosis (Gardner & Eberhart, 1966) and studying the genetic control of evaluated characters (Hayman, 1954).
There are also modifications adapted to partial diallels, such as Geraldi & Miranda Filho (1988) and Miranda Filho & Geraldi (1984), adapted from Griffing (1956) and Gardner & Eberhart (1966), respectively. Partial diallels were developed to increase the number of parents included in diallel crosses (Kempthorne & Curnow, 1961;Hallauer et al., 2010) and have provided information on the presence and magnitude of additive, non-additive and heterosis gene effects, without the need of crossings between all parents (Cruz et al., 2012).
The present work aimed to estimate the combining ability (general and specific) and heterosis manifested in experimental hybrids obtained from diallel crosses between groups of eggplant lineages with agronomic potential, under high temperature conditions, in order to identify promising combinations for the selection of superior genotypes or that are competitive with the currently available hybrid cultivars.

MATERIAL AND METHODS
The experiment was conducted on Universidade Federal Rural de Pernambuco (UFRPE), Recife, Pernambuco, from April to December 2017.
In order to obtain F1 hybrids, 7 parents from the Embrapa Hortaliças Eggplant Germplasm Bank and previously selected (Valadares et al., 2019) were crossed in a 3x4 partial diallel arrangement, totaling twelve hybrid combinations.
The parents were stratified into two groups. Group 1 consisted of the genitors CNPH 135, CNPH109 and CNPH 47 and group 2 consisted of the genitors CNPH 141, CNPH 60, CNPH 53 and CNPH 51. Crosses were performed manually, emasculating flowers of the female genitors (group 2) and proceeding to pollination with pollen extracted from male genitors (group 1). Before and after crosses, measures were taken to ensure the genetic identity of each cross.
The plants were cultivated in open hydroponic system with substrate in a greenhouse. Mineral nutrition and water requirement were supplied by balanced nutrient solution at each stage of plant development, by 2 L h -1 drip irrigation system, automatically controlled by a digital timer, with irrigation amounts and duration adjusted according to weather conditions in the region and amount of nutrient solution absorbed daily by the plants.
The experimental design was a complete randomized block with 20 treatments (genotypes), twelve hybrid combinations obtained by partial diallel 3x4, seven genitors' lineages and one hybrid Ciça F1 as control and four replications.
The plots consisted of four pots filled with substrate (coconut powder) containing one plant each, spacing 1.75 m between rows and 0.60 m between plants.
The average data of each trait was subjected to analysis of variance (p<0.01) and averages grouped by Scott-Knott procedure (p<0.01). To obtain general (GCA) and specific (SCA) ability estimates, the F1's genitors and hybrids averages were submitted to diallel analysis according to the partial diallel model proposed by Geraldi & Miranda Filho (1988), adapted from the model 2 proposed by Griffing (1956). Estimates of heterosis relative to the parental averages were obtained for all hybrid combinations by the equation Hr = [F̅ 1/(P̅ 1+P̅ 2/2)x100] for each F1 hybrid combination. Analyzes were performed using the GENES program version 1990.2018.75.

RESULTS AND DISCUSSION
The micrometeorological data obtained during the experiment period showed that the maximum air temperature in greenhouse ranged between 38.2 and 52.5ºC and the minimum temperature between 19.6 and 24.9°C. The average temperature ranged from 27.8 to 33.5ºC. Thus, the environment was classified as high temperature for eggplant cultivation. Relative humidity ranged from 65 to 79%.
Decomposing the effects of treatments (genotypes) on general combining ability (GCA) and specific combining ability (SCA), significant differences for GCA were observed between group I and group II parents for most traits, except for pollen viability (PV). This indicates that the parents of groups I and II are heterogeneous and that the action of additive gene effects was able to influence the expression of the traits (Table 1).
The effects of SCA were significant for most traits except fruit length (FL) and fruit width (FWi). Thus, except these two characters, non-additive gene effects act and were important in the genetic control of traits. However, the upper squared averages of the GCA indicate that additive gene effects were more important in trait genetic control, except for PV, for which the nonadditive effect was higher (MSSCA>MSGCA) ( Table 1).
In group I, only genitors CNPH 135 and CNPH 109 had positive GCA estimates for fruit fixation index (FFI), number of fruits per plant (NFP) and production per plant (PP), although negative for PV and fruit weight (FWe). In group II, genitor CNPH 51 stood out with positive GCA for most traits, except for FWe and FWi. High estimates of GCA, whether positive or negative, indicate that the genitor is better or worse than the others with respect to the average behavior of crosses (Griffing, 1956) and is preferred to constitute new populations (Miranda Filho et al., 1988).
The genitors' GCA estimates were both positive and negative for fruit weight (FWe). The genitors CNPH 47,  (Table 1). These results are important because they indicate that the parents have both favorable and unfavorable effects for these traits and can be used in crossings aiming at different fruit shapes and sizes.
Only the genitor CNPH 141 presented negative values for the characters FWe, FL, FWi and FLWR indicating that its use in crossings will not result in superior progenies (Table  1). When GCA is high and negative, it demonstrates that a given genitor is inferior to the other diallel parents (Cruz et al., 2012).
Positive and/or negative GCA estimates were observed for plant height (PH). In this case, most genitors had estimates close to zero ( Table 1). Estimates of near-zero GCA indicate that the genitor does not differ from the overall average of crossings (Griffing, 1956).
Regarding the effects of SCA, the 1x4, 1x7 and 2x5 hybrids showed nonadditive gene effects for FFI, NFP and PP, although negative for FWe, due to the negative genetic correlation between FWe and the other traits (Valadares et al., 2019) (Table 1).
Among these, the 2x5 hybrid also presented positive SCA for PV, FL, FWi and FLWR. The best hybrid combinations have high SCA effects, positive or negative, and crosses should occur between divergent genitors, in which at least one of them has high GCA (Griffing, 1956;Cruz & Vencovsky, 1989). The 1x4 and 3x4 hybrids showed positive estimates for both GCA and SCA for FFI, NFP and PP, showing higher potential for crossbreeding to obtain genotypes with high fruit fixation under high temperatures (Table 1).
For FL, FWi and FLWR, the estimates for SCA were positive for most hybrids (Table 1). For PH the SCA estimates were positive and/or negative, but close to zero, indicating that the performance of the hybrids occurred as expected in relation to their parents (Table 1). The 1x4, 1x5 and 1x6 hybrids expressed positive heterosis for most traits analyzed (Table 2). Positive, as well as negative heterosis in eggplant hybrids are reported for these traits by several authors (Shafeeq et al., 2007;São & Mehta, 2010;Dharwad et al., 2011;Singh et al., 2012Singh et al., , 2016Kumar et al., 2013;Dubeyet et al., 2014;Reddy & Patel, 2014;Magar et al., 2016;Sivakumar et al., 2017;Patel et al., 2017).
Overall hybrids had higher averages than parents for most characters (Table  2). However, only 1x4 hybrid stood out with the highest averages for PV, FFI, NFP and PP (Table 2). FFI is the trait most influenced by high temperatures in the Northeast Brazil, because under high temperatures it is drastically reduced, beside that it has a genetic correlation with the traits NFP and PP (Valadares et al., 2019).
Although the percentages of PV, FFI, NFP and PP were close to or below the values reported by other authors (Silva et al., 1999;Baswana et al., 2006;Valadares et al., 2019), the averages obtained for hybrids and genitors were superior to those obtained for Ciça in most traits, except for FWe, FL and PH, where Ciça stood out among the best genotypes (Table 3).
For FLWR, indicative of fruit shape, genitors and hybrids differed in the vast majority from the Ciça hybrid (Table  3). In this case, because genotypes with variation in fruit size and shape were used, the variability observed in the crosses indicates the possibility of selection of fruits with variations in shape and size, and not exclusively similar to the hybrid Ciça.
In addition, FLWR values should not be considered isolated to avoid misclassification of fruits, even if genotypes produce fruits with desired shape. It is therefore recommended that the FLWR values be analyzed in conjunction with FL and FWi values. So, for FWi, some hybrids and genitors did not differ from the commercial hybrid Ciça (Table 3), while for FL all differed from Ciça. Variation in FL and FWi have been reported by other authors (Sao & Mehta, 2010;Singh et al., 2012;Kumar et al., 2013;Dubey et al., 2014;Sivakumar et al., 2017;Valadares et al., 2019).
Additive gene effects were more important than nonadditive ones for most analyzed traits, except for PV. Genitors CNPH 141, CNPH 135, CNPH 109 and CNPH 51 stood out with favorable gene effects to obtain high temperature tolerant genotypes, having good GCA for FFI, NFP and PP. The 1x4 and 3x4 hybrids showed positive estimates for both GCA and SCA, showing higher potential for crossbreeding to increase FFI, NFP and PP under high temperatures.
The 1x4, 1x5 and 1x6 hybrids expressed positive heterosis for most analyzed traits.
The 1x4 hybrid stood out with the highest averages for PV, FFI, NFP and PP.
For FWe, FL, FWi and FLWR, both positive and negative heterosis were observed, as a consequence of the phenotypic variability of the genitors for these traits and suggests the possibility of selection for different fruit formats and sizes.