Selection of loose-leaf lettuce breeding lines based on non-parametric indexes

Non-parametric selection indexes (that is, require no estimation of parameters) can be used to help the selection process in the final stage of genetic enhancement. In this context, the objective of this study was to evaluate the efficiency of non-parametric selection indexes to choose promising lineages of loose-leaf lettuce, so that genetic gain of each lineage can be estimated and then registered in the Ministry of Agriculture. The indexes of Mulamba and Mock, Elston and Schwarzbach were used in the analysis. Ten genotypes of loose-leaf lettuce were evaluated, with eight of them being lineages (L1, L2, L3, L4, L5, L6, L7, L8) and two commercial cultivars (Vanda and Vera). The experiments occurred in six different locations of cultivation during the autumn and winter seasons of 2014. Experimental delineation was composed of random blocks, with four repetitions and the evaluated characteristics were: total production, commercial production, number of leaves, plant volume and stem length. The effects of genotype (G), location (L), and G x L interaction were significant for all characteristics. Indexes were correlated to classify the genotype. The Mulamba and Mock index stood out because it enabled good direct gains for the evaluated characteristics and because of its easy construction. Therefore, this index is recommended for selection of loose-leaf lettuce genotypes in the stage of cultivar recommendation in different locations. As the best lineages of loose-leaf lettuce were L2, L3, L7 and L8, these lineages have been considered promising and are recommended for registration.


INTRODUCTION
Lettuce is one of the most consumed vegetables in the world and its consumption has grown yearly.This is mainly due to new dietary habits that the population has adopted, which includes lettuce as an indispensable ingredient of its meals.In this context, market demand for high quality lettuce is already a reality, since consumers have become more selective and critical when choosing their foods.The Lettuce Breeding Program of the University of the State of São Paulo (UNESP-FCAV) has been active since 2003 and has worked on two different matters.The first regards Bremia lactucae L. breed identification in lettuce production regions of the State of São Paulo involves annual monitoring (Braz et al., 2007;Castoldi et al., 2012;Galatti et al., 2012;Nunes et al., 2016;Souza et al., 2011).The second is related to crossing of loose-leaf lettuce genitors and differentiating lettuces of Dm genes resistant to B. lactucae, with the intention of obtaining new lineages with positive agronomic and B. lactucae resistant traits (Castoldi et al., 2014).Currently, the program works on promising lineages that can be evaluated in different cultivation locations.
Agronomic yield assays performed in different locations which use different cultivation systems is an important stage of genetic enhancement, because necessary agronomic and productive characteristics are tested so that new lineages can be recommended in the future.The use of non-parametric selection indexes (indexes that do not require parameter estimations) can help in the lineage selection process when getting to the final stage of genetic enhancement.Non-parametric indexes have the intention of classifying genotypes in a simple way (Garcia and Souza Junior, 1999).
A few studies have reported the efficiency of nonparametric selection indexes in the beginning of progenies enhancement programs (Neves et al., 2011;Oliveira et al., 2008), as well as in advanced lineage selection stages (Marinho et al., 2014;Vittorazzi et al., 2013).However, studies using selection indexes on lettuce are still scarce.In this context, the objective of this study was to verify the efficiency of non-parametric selection indexes in choosing promising lineages of loose-leaf lettuce in order to register them in the Brazilian Ministry of Agriculture and to estimate genetic gains based on the same indexes.

MATERIALS AND METHODS
The experiments were installed during autumn and winter of 2014, in the cities of Monte Alto, São Simão, Aramina, Mogi das Cruzes, Biritiba Mirim, and Salesópolis -São Paulo State (Brazil).Experimental design was based on random blocks with six experimental groups and four repetitions.Treatments composed of 10 genotypes (8 lineages and the 2 commercial cultivars Vanda and Vera) and six (6) locations (Monte Alto, São Simão, Aramina, Mogi-das-Cruzes, Biritiba-Mirim, and Salesópolis).Each plot composed of 28 plants distributed in 4 rows with each row 1.75 m long and plants were spaced at 0.25 m intervals, but only the 6 central plants of each group were evaluated.The evaluated lineages came from the initial crossings of JAB 4-13-7 (male genitor, possessor of the DM18 gene of B lactucae resistance) with commercial cultivars: Argelis (A)possessor of the B. lactucae resistance factor R-38, Vanda (V), Venerada (Vn) and Solaris (S) (all female genitors).This lineage resisted the B. lactucae breeds that occurred in the State of São Paulo (Castoldi et al., 2014).
The agronomic evaluated characteristics were: total production in g plant -1 (TP), commercial production in g plant -1 (CP), number of leaves per plant (NL), volume of the plants (cm 3 planta -1 ) (V) and stem length in cm (SL).These measurements were done at harvest point (45 days after transplantation).Total production (TP) was obtained by calculating the average of the fresh mass in 6 plants (without extracting the old leaves).Commercial production (CP) was obtained by measuring the average weight of the fresh mass in 6 plants after removal of old leaves and of spare stem.NL represents the total number of leaves that reached the length of at least 3 cm in each plant.Volume (VOL) was calculated by multiplying the diameters of extremities, D1 and D2, and the height (h) of 6 plants from each of the portions (groups).In order to calculate V, the volume formula of 2 diameter used ellipsoid was: V = 4/3 π (D1/2) (D2/2) h.Stem length (SL) was measured using a graduated ruler.The obtained data were initially submitted to the normality test and homogeneity of residual variances, and normality of the data was detected.Subsequently, data were submitted to statistical analysis of variance by location (environment), in which the variation sources were considered random.To do so, the statistical genes computer program (Cruz, 2013) was used.The formula adopted for joint analysis was: Where, Yijk is the mean phenotypic value of the group; m is the average; gi is the fixed effect of the i -th genotype; b/ajk is the effect of k repetition in location j; gaij is the interaction effect of genotype i in location j and eijk is the experimental error.The selection indexes were constructed following Mulamba and Mock (1978), Elston (1963) and Schwarzbach (1972) and Wricke and Weber (1986).The expected gain by direct selection in the i -th character was estimated based on selection indexes studied through the following expression: Where, is the average of selected samples for characteristic i; is the original average of the population; is the selection differential carried out in the population; and is the heritability of characteristic i.In order to compare lineage classification using the indexes of Mulamba and Mock (1978), Elston (1963) and Schwarzbach (1972), the Spearman correlation between them was obtained.

RESULTS AND DISCUSSION
Significant effects were verified for genotypes, locations and genotype x location (GXA) for all evaluated characteristics (Table 1).For the Monte Alto location (Table 2), the Mulamba and Mock (1978) indexes selected lineages L 2 , L 3 and L 7 , whereas the Elston index (1963) indicated lineages L 2 , L 3 and L 4 and the Schwarzbach index (1972) chose lineages L 4 , L 7 and L 8 .Lineages L 2 and L 3 were indicated by both the I MM and I E indexes.Index I MM presented the lowest rank; D i represented shortest distance from the genotype to the ideotype and I E the highest value.Lineage L 4 indicated by indexes I E and D i , also presented higher results for TP, CP and V.These characteristics are relevant for lettuce cultures since the consumer market demands bulky plants and good-looking leaves.Thus, lineages L 2 , L 3 and  The other lineages and commercial cultivars Vanda and Vera did not present satisfactory agronomic development.Their selection index results were low when compared to the lineages selected, which does not justify their cultivation in the Monte Alto region.Direct gains in Monte Alto, based on the adopted 30% selection intensity, were higher for V, with a 14.98 to 16.17% increase by the I MM and I E indexes.Additionally, higher total gains were obtained in all characteristics by the same indexes.The D i index showed no satisfactory gains when compared to indexes I MM and I E , especially for TP and V (Table 3).Rosado et al. (2012) found different results than the ones presented in the current study where they worked with simultaneous selection using selection indexes in progenies of sour passion fruit trees and verified the use of the Elston index which was not capable of distributing gains fitting for the purposes of that research.Thus, based on the results, it is possible to deduce that gains from index are subject to variation according to the greatness of the evaluated measurements.However, the same authors found that the index based on the additions of the Mulamba and Mock rankings was the most adequate, because it promoted balanced gain distribution.This result did support the results obtained in the current study.
In the region of São Simão (Table 2), indexes did not indicate the same lineages as the ones selected in Monte Alto.The Mulamba and Mock (I MM ) index identified 5 lineages: L 2 , L 3 , L 4 , L 7 and L 8 .On the other hand, the Elston index (I E ) indicated lineages L 4 , L 7 and L 8 , which was in agreement with the other 2 indexes that selected lineages L 2 , L 7 and L 8 .The Schwarz back index (D i ) also indicated lineages L 4 , L 7 and L 8 for plantation in São Simão, which is also in agreement with the I MM index.
The highest direct gain was in NL (9.17%), measured by D i , however the highest gain values were for the I E

Table 1 .
Variation analysis chart of 6 random block experiments with ten loose-leaf lettuce genotypes, conducted in 6 different locations in autumn-winter, 2014.

Table 2 .
Indexes: Mulamba and Mock (IMM), Elston (IE) and Schwarzback (Di), applied to the characteristics total production (g/plant); commercial production (g/plant); number of leaves; plant volume (cm³/plant) and stem length (cm) for 10 genotypes of loose-leaf lettuce in 6 different locations in the autumn and winter seasons of 2014.

Table 3 .
Estimates of genetic gains in percentage, by the indexes of selection of Mulamba and Mock (IMM), Elston (IE) and Schwarzback (Di) applied to the characters total production (TP) (g / plant); commercial production (CP) (g / plant); number of leaves (NL); plant volume (V) (cm³ / plant); stem length (SL) (cm) for ten genotypes of crisphead lettuce, in six environments, in the fall-winter of 2014, based on the selection at each site..