Introduction

In Mexico maize (Zea mays L.) is the most important crop, concentrating 7.5 million hectares and is the basic diet component of most Mexicans. However, in the country only 22.1 million tons of corn are produced each year and 10 million tons are imported to cover the apparent demand (^{Tadeo et al., 2015a}). Given that maize is the mainstay of Mexican food, what happens to this crop has socioeconomic repercussions in the country, since its main derivative, the tortilla, is the staple and preponderant food of the population.

In 2013, 2 million hectares of maize (26.6% of the national area planted with this species) were planted in the Valles Altos region of Mexico (2 200 masl) with an average yield of 2.7 t ha^-1 (^{SAGARPA, 2013}). Based on the magnitude of this average yield, it can be said that the production per unit area is low and it is necessary to increase it, because as time goes by the crop area is getting smaller but the demand continues to increase (^{Virgen et al., 2016}).

In addition to low maize production, in Mexico there are 31 million people with malnutrition, of which more than 50% are severely affected (indigenous and low-income urban population) (^{Sierra et al., 2010}). Most cereals, including maize, staple food in Mexico, have low protein content and the availability of essential amino acids such as lysine and tryptophan is limited. However, to counter malnutrition, there are maize with quality of protein, called QPM for quality protein maize. These maize were derived from the use of the mutant opaque-2 (opaco-2) gene from Peru (^{Mertz et al., 1964}), which in homozygous recessive condition (^{Mertz, 1994}) expresses the same total amount of proteins, but with contents of up to 100% more lysine and tryptophan than the common maize, but due to its floury texture, grain weight and field yield are low, as well as being easily attacked by pests (^{Mertz, 1994}; ^{Espinosa et al., 2005}).

^{Vasal et al. (1980)} using conventional breeding techniques succeeded in incorporating modifying genes into opaque-2 maize, which produced a similar texture in the opaque endosperm as normal maize, but with high protein quality in the grain. Thus, in the 1980s protein quality maize (QPM) was obtained without the initial disadvantages, with a harder grain texture than opaque maize and with common or normal appearance (^{Vasal, 2001}).

The National Institute for Forestry, Agriculture and Livestock Research (INIFAP), in coordination with the International Center for Maize and Wheat Improvement (CIMMYT), carried out eight research works with QPM in different regions of Mexico with the participation of more than 60 specialists, whose lines were: genetic improvement, seed and production technology, forages, stored grain pests, agronomic management, technology transfer, effect on human diet and animal diet and quality analysis. As a result of these activities, more than 30 hybrids and varieties were registered in the catalog of varieties feasible of certification (CVC), now national catalog of plant varieties (CNVV) (^{Espinosa et al., 2005}).

On the other hand, commercial seed production of maize hybrids requires to properly remove the spikes of the female line to obtain high quality seeds with genetic identity. This is a manual activity and involves 24 to 50 wages per ha^-1. Which increases production costs (^{Tadeo et al., 2014a}, ^{Tadeo et al., 2014b}). About this the use of androesterility may be a viable option for obtaining hybrid seeds of good genetic quality but cheaper (^{Tadeo et al., 2014b}). The use of different types and sources of gene-cytoplasmic male sterility is an opportunity for commercial production of hybrids, since the use of andro-sterile lines as female progenitors (^{Stamp et al., 2000}) reduces the costs and effort that taking out spikes represents.

In Mexico, lines, varieties and hybrids (simple and trilineal crosses) of maize have been developed in the Facultad de Estudios Superiores Cuautitlán (FESC-UNAM) (^{Tadeo et al., 2014a)}. In the case of hybrid progenitor lines, there are versions that have the andro-sterile characteristic, to which also the opaco-2 gene has been incorporated to obtain high quality of protein (QPM) (^{Espinosa et al., 2003}).

At Valles Altos of México, improved varieties and hybrids of maize with high protein quality are not commercially planted (^{Espinosa et al., 2003}), so the objective of this research was to determine the productivity of 18 simple crosses btween six androthermal lines and three elite lines, all of them QPM, with the purpose of defining the perspective of these lines and their probable advantage by their integration to maize genetic improvement programs. It was hypothesized that when evaluating the materials in two localities of Valles Altos, there would br differences between the simple QPM crosses for their agronomic response, and that there would also be differences between them and the two normal quality commercial trilineal hybrids used as controls.

Materials and methods

In this paper the evaluated materials were 18 simple hybrid of QPM maize and the trilineal hybrids H53 AE and TSIRI PUMA of normal grain as controls. Simple hybrids are crosses between 6 QPM andro-sterile lines (LAEQ1, LAEQ2, LAEQ3, LAEQ4, LAEQ5 and LAEQ6) and 3 elite QPM lines (CML173, CML354 and CML352). The experiments were established in the spring-summer 2014 cycle in two locations. The first one, located in the municipality of Cuautitlán Izcalli, Estado de México (19° 41’ north latitude, 99° 11’ west longitude, 2 274 m altitude), whose soil is of loamy. The other corresponded to Santa Lucía de Prías, Coatlinchán, Municipality of Texcoco, Estado de México (19º 27’ north latitude, 98º 51’ west longitude, 2 240 m altitude), with sandy loam textured soil.

In both locations genotypes were distributed in a randomized complete block design with three replicates. A population density of 65 000 ha^-1 plants was used, where the experimental plot consisted of a furrow 5 m long and 0.8 m wide.

The ground preparation consisted of a step of plow, two of drag, leveling and furrow. The planting took place in June 2014 in both localities. In Santa Lucia, irrigation was applied to the sowing and two irrigation schemes were applied. In Cuautitlán irrigation was only applied to the sowing, then the cycle had rainfall. The harvest in the two localities was made in December of 2014. It is worth mentioning that during the crop cycle in Santa Lucia two hailstorms occurred before the male flowering. The climatic data (maximum and minimum temperature, as well as rainfall) were obtained from the meteorological stations, Almaraz of the FESC-UNAM and the Universidad Autónoma de Chapingo (Figure 1).

Figure 1 Maximum and minimum monthly temperature, and monthly precipitation during the crop cycle of 20 hybrids in two localities of Valles Altos, 2014. 

The recorded rainfall from sowing to harvest was 825 mm in Cuautitlán, and 723 mm in Santa Lucia, rainfall that was unevenly distributed (Figure 1). The maximum and minimum temperature variation in both environments was very low (0.5 to 1 °C). The maximum temperature in Cuautitlán occurred in August, reaching 26.4 °C, while in Santa Lucia the maximum temperature (25 °C) occurred in June and August, coinciding with sowing and beginning of flowering, respectively. Also in Santa Lucia hail occurred in the first and second weeks of August, so it affected the crop in previous days and during the beginning of flowering. However, this setback the plants had a good leaf area, allowing the crop to recover but with adverse effects.

The studied variables were male flowering (FM, days), when 50% of the speaks per furrow appeared; plant height (AP, cm) and cob height (AM, cm) in 5 to 10 plants chosen randomly, the distance from the base of the plant to the apex of the spike (AP) and from the base of the plant up to the upper cob knot (AM) was measured. During harvesting, in another 5 to 10 plants, the field weight (PC, kg) was taken and also the total weight of the cobs. Then on each cob the number of rows (HM) and the number of grains per row (GH) were counted. The number of grains per cob (GM) was the product of the average rows of each cob per average grain per row. Subsequently, the cobs were harvested and the percentage of grain moisture was determined in a Stenlite determinant (Burrows^® brand, model 700). The grain yield (REND) was obtained with the formula:

Where: PC= field weight of the harvested cobs (kg) of each plot; MS= dry matter percentage, obtained by the difference of 100 minus the percentage of moisture obtained from the Stenlite apparatus; PG= grain percentage, which is the ratio average of the grain weight to the cob weight without bracts (five cobs), multiplied by 100; FC= conversion factor, which is obtained by dividing 10 000 m² between the size of the useful plot in m²; 8 600= constant to estimate the yield with a commercial humidity of 14%.

A comparison test of means (Tukey p= 0.05) was made for each variable; both procedures were performed using the SAS package (^{SAS Institute, 2002}).

Results and discussion

The combined analysis of variance (Table 1) detected high significance (p≤0.01) between environments (A), between genotypes (G), and in interaction (AG) for all variables, which means that environmental conditions and effects were different in this evaluation. The coefficient of variation ranged from 1% (FM) to 9% (REND), indicating an acceptable control (<20%) of the experimental variability. These results suggest that there is genetic variation between crosses which is useful in a breeding program for QPM maize by hybridization.

Table 3 Mean squares and statistical significance of the combined analysis for several variables of simple crosses of maize in two environments. Cycle spring-summer 2014. 

^**= significancia estadística al 0.01 de probabilidad; ^*= significancia estadística al 0.05 de probabilidad; FV= factor de variación; GL= grados de libertad; REND=rendimiento de grano; FM= floración masculina; AP= altura de planta; AM= altura de mazorca; LM= longitud de mazorca; HM= hileras de la mazorca; GH= granos por hilera; GM= granos por mazorca; CV= coeficiente de variación.

Regarding to the response of the simple crosses in average by environments, the yield in Cuautitlán presented an average of 7 000 kg ha^-1, being superior (p≤ 0.05) to that of Santa Lucia, whose average was 4 800 kg ha^-1. The grain yield in Santa Lucía was reduced 30% due to the occurrence of two hailstones that caused damage to the photosynthetically active leaf area. In addition, there was a difference of 100 mm of precipitation between both localities, being this in favor of Cuautitlán.

The maximum yield in average of the two localities corresponded to the crossing LAEQ5 CML354 (8 780 kg ha^-1), while the smaller yield was for the cross LAEQ1 CML352, with 3 160 kg ha^-1 (Figure 2). In Cuautitlán, the crosses with the highest average yield were LAEQ5 CML354, LAEQ4CML173 and LAEQ3 CML173, whose yield was 9 340 kg ha^-1, being a very similar yield (9.3 t ha^-1) to that obtained by ^{Torres et al. (2011)} in an evaluation of trilineal hybrids. Fort he aforementioned, these more yielding crosses in Cuautitlán are recognized as highly promising materials for plant breeding purposes, when used as females in the formation of trilineal hybrids. In Santa Lucia the crosses with higher average yield were also LAEQ5 CML354, LAQ4 CML173 and LAEQ3 CML173 with yields close to 8 000 kg ha^-1. The yields obtained in Cuautitlán demonstrate the potential value of these crosses in the spring-summer cycle, because in this environment there were no unfavorable environmental conditions that affected the crop, like the presence of hail on two occasions in Santa Lucía.

Figure 2 Grain yield of single maize crosses and two trilineal hybrids, in two environments. Cycle spring-summer 2014. Averages with the same letter inside each bar are statistically equal (Tukey p= 0.05, DMS= 1 150).  

50% of the single crosses had an average experimental yield superior to that of the controls and to the general average (5 875 kg ha^-1), being the reason why these crosses are considered of good productivity, because they produced yields similar to those of commercial hybrids such as H40, H44, H52, H66, H70 (hybrids developed by INIFAP, registered with the National Catalog of Varieties of Plants (CNVV) in 1999, 1998, 2006, 2009 and 2010 respectively) and BUHO (developed by Asgrow and registered with the CNVV in 2006), which have yields ranging from 5.5 to 8.9 t ha^-1 (^{Rivas et al., 2011}).

On the other hand, a similar yield was observed between the crosses that were located in the first groups of significance of each locality, because in their formation the same male progenitors (elite lines CML173 and CML354) intervened, each participating four times in the best crosses according to their yield). It was also observed that the control trilineal hybrids presented lower values than the general average, and therefore half of the single crosses were superior to them, which again shows the importance of the evaluated material in QPM maize breeding programs by hybridization.

Male flowering presented a statistical difference between environments, with the genotypes being the later ones for eight days (87 days) in Cuautitlán with respect to the genotypes in Santa Lucia (79 days). In Cuautitlán, male flowering began at 81 days in some crosses, and the later ones started a week later; the TSIRI PUMA trilineal hybrid was the earliest genotype (80 days). The crosses that had the elite line CML173 as male were the later, from 87 to 90 days, and it was found that 55% of the simple crosses presented late bloom (Figure 3). This is explained by the fact that the line CML173 is of subtropical origin and therefore late (^{Sierra et al., 2004}; Sierra et al., 2010).

Figure 3 Male flowering (FM) of simple maize crosses and two trilineal hybrids, in two evaluation environments. Cycle spring-summer 2014. Averages with the same letter inside each bar are statistically equal (Tukey p= 0.05, DMS= 2.3). 

In Santa Lucia, male flowering of the materials occurred between 76 and 82 days. The hybrid TSIRI PUMA and the crosses LAEQ6 CML352 and LAEQ3 CML352 showed higher precocity (76 days), while some crosses began their flowering after day 78. In another similar study with different materials, including fertile androsterile hybrids, presented 78 days to male flowering (^{Tadeo et al., 2015}), so some of the materials evaluated in this study are considered to be of acceptable precocity, for irrigation tip and good temporal sowing, understanding that irrigation tip sowing is that in wich the soil has moisture once it has been watered, subsequently sowing and water needs throughout the development of the plant are covered with rainfall. The most recent single cross was LAEQ3 CML173, whose anthesis occurred at 82 days. In both localities, 17% of the crosses were early, the rest are considered as late genotypes.

In this paper, the lines used as male progenitor have 30 to 40% of temperate germplasm and 60 to 70% of tropical germplasm. In this regard, ^{Lafitte (2001)} mentions that most of the tropical cultivars are sensitive to photoperiod and that the extent of this sensitivity varies from one to 12 days of anthesis delay. This helps to explain the difference in the number of days of anthesis between environments (from 4 to 11 days), together with the difference of precipitation between the two localities and the damage to the plants caused by hail in Santa Lucía de Prías.

The plant height presented a statistical difference between environments, being the genotypes shorter in Cuautitlán (207 cm) than in Santa Lucía (249 cm). Similar results were found in ^{Tadeo et al. (2014b)}, where also the genotypes in the environment of Santa Lucia were higher. In relation to this, if it is intended to make use of these simple crosses hybrids within a hybridization program, plant height is a very important variable, because the height influences the activities of spiking and pollination; thus, plants with lower height are more easily manipulated at the time of taking out the spikes, which allows to perform the activity in a shorter time in comparison with plants of higher height (^{Virgen et al., 2014}).

The average plant height in Cuautitlán varied from 163 to 230 cm, being these values lower than those presented in Santa Lucía because there were areas with flooding within the plot. In Santa Lucía, the average height varied from 200 to 280 cm. On the other hand, 56% of the simple crosses presented differences between environments of 29 to 88 cm, which would allow to select those with height of interest. Regarding the witnesses, the hybrid H53 AE had heights of 192 to 280 cm, and their difference between environments was of 88 cm, whereas the hybrid TSIRI PUMA showed heights of 198 to 239, with a difference between environments of 41 cm.

It should be noted that two crosses whose male progenitor was the elite line CML173, as well as three crosses with the line LAEQ2 as female progenitor, were crosses that expressed lower difference of height between environments. For example, the LAEQ2 cross CML173 expressed in Cuautitlán 224 cm and in Santa Lucía 234 cm, while the LAEQ2 cross CML173 showed heights of 222 to 250 cm between environments. This would allow to select consistent crosses with its average plant height value across environments.

The phenotypic correlation coefficient between grain yield and agronomic variables and yield components allowed to identify a positive association between cob length (LM), grains per row (GH) and cob diameter (DM) with yield (REND) (p≤ 0.01) (Table 2). The highest and most significant correlation coefficients were: FM-FF (0.99), FM-LM (0.76), LM-GH (0.76), LM-REND (0.74), GM-REND (0.67), DM- FM (0.67), DM-FF (0.68), DM-REND (0.73), GM-GH (0.8). These correlations are important since these traits directly influence maize yield and are frequently used by researchers and peasants to select genotypes of their interest.

Table 2 Pearson correlation between several variables of simple maize crosses in two environments. Cycle spring-summer 2014. 

^**= significancia estadística al 0.01 de probabilidad; ^*= significancia estadística al 0.05 de probabilidad; REND= rendimiento de grano; FM= floración masculina; FF= floración femenina; LM= longitud de mazorca; GH= granos por hilera; DM= diámetro de mazorca; GM= granos por mazorca.

Conclusions

The best simple crosses of quality protein maize (QPM) among localities were LAEQ5 CML354, LAEQ4 CML173 and LAEQ3 CML173, with yields greater than 8 t ha^-1, surpassing the commercial controls in 43 to 57%. This positions them to participate in the formation of superior trilineal QPM hybrids, combining them with a third line that generates a good hybrid.

By their participation in the best crosses, the outstanding QPM androesterile experimental lines were LAEQ3, LAEQ4, and LAEQ5, since each participated twice in the 9 highest performance crosses. The best QPM elite lines were CML 173 and CML 354.

83% of the genotypes were late in this research, beginning anthesis after 82 days; however, this flowering is classified as intermediate within the values for this variable in Valles Altos, which will allow its use in the conformation of trilineal hybrids to be used in irrigated or good temporal sowing. Genotypes with acceptable plant height for seed production were also identified. On the other hand, the characteristics that were highly related to the yield of the QPM maize simple crosses of this study were the days to flowering and the cob’s length and diameter.