A comparison of controlled self-pollination and open pollination results based on maize grain quality

Maize (Zea mays L.) grain endosperm is triploid (3n), of which 2n come from the male (transferred by pollen) and only 1n from the female plant, thus a major impact of the male form can be expected on grain quality parameters. A good example of this relationship is the phenomenon of xenia. The aim of this study was to determine the effect of pollen on grain quality. The f ield experiment was conducted in 2011; seeds were harvested from eight cultivars: Bosman, Blask, Tur, Kozak, Bielik, Smok, SMH 220 and Kresowiak, derived from free pollination and controlled selfpollination of maize. Analyses of nutrient contents and starch content in the grain were conducted in the laboratory. In addition, 1,000 grain weight and the hectoliter weight of all grain samples were recorded. The results confirmed differences in grain quality of maize hybrids obtained by self-pollination and by open pollination. Grain of maize plants obtained by open-pollination was characterised by higher contents of N-free extract and starch, and lower protein content. Undertaking further studies on this subject may indicate specific recommendations for agricultural practice, such as mixtures of hybrids with good combining abilities, which will contribute to improved grain quality without additional costs. Additional key words: chemical composition; hybrids; pollen; pollination. * Corresponding author: ptasz@up.poznan.pl Received: 09-09-13. Accepted: 13-05-14. This paper has one Supplementary Figure that does not appear in the printed article, but accompanies the paper online. Abbreviations used: CV (coefficient of variation); DM (dry matter); NFE (N-free extract); PCA (principal component analysis); S (starch); SC (two way cross-breed); TC (three way cross-breed); TGW (thousand grain weight) Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA) Spanish Journal of Agricultural Research 2014 12(2): 492-500 http://dx.doi.org/10.5424/sjar/2014122-4970 ISSN: 1695-971-X eISSN: 2171-9292 RESEARCH ARTICLE OPEN ACCESS increase between 1939 and 2004 was approximately 100 kg ha · yr, i.e. 2% per year (Tollenar & Lee, 2006). Currently maize hybrids up to the so called FAO 300, grown in Polish agriculture, come from leading breeders, both domestic and from the EU and the USA. In experiments on the assessment of individual hybrids, maize plants are sown close together. Maize plants as cross and wind pollinated species can be pollinated with pollen from considerable distances, according to Felsot (2002) from up to 300 meters. However, experiments with coexistence of conventional and genetically modified maize showed that the main pollination is done within a much shorter distance, up to 20 m (Brookes et al., 2004; Arritt et al., 2007). Grain endosperm is triploid (3n), of which 2n come from the male (transferred by pollen) and only 1n from the female plant, so a major impact of the male form can be expected on grain quality parameters. A good example of this relationship is the phenomenon of xenia observed on sweet maize pollinated by fodder maize. The xenia effect in endosperm and embryo traits has been known for a long time (Davarynejad et al., 1994; Seka et al., 1995; Bulant et al., 2000; David et al., 2001). Maize grain is harvested from the F2 generation of the grown hybrid, but only if it is pollinated with the pollen of the same cultivar, as happens in large fields. For small plots used in variety trials, free open pollination by pollen from varieties sown in close proximity is observed and the resulting kernels are hybrids from the neighbouring varieties. In the case of twoway hybrids grown, random four-way hybrids are obtained, which may show some effects of heterosis as well as the effect of splitting of the traits. Thus, grain from such pollination may differ in its chemical composition and other characteristics from that obtained from self-pollination of only one variety. The results of this evaluation will not be comparable with those collected from large f ields. As it was observed by Mercer et al. (2002), there is a preference of other plant pollen (allopollen) over own pollen (autopollen) during fertilization of maize plants, with an average rate of 73.27% in the ear, which is close to the average frequency of 80.06% observed also by Balestre et al. (2007a). The aim of this study was to determine the effect of pollen on grain quality. The research hypothesis assumes that maize grain produced from free pollination with pollen of different varieties grown next to each other on side-by-side plots varies in its chemical composition and other characteristics from that produced by controlled self-pollination. Material and methods Plant material and experimental design Field experiment was conducted at the Plant Breeding Station in Smolice (Poland) in 2011. Seeds were harvested from eight cultivars diverse as regards of the length vegetation period expressed according to FAO classification: Bosman (C1, FAO 250; medium-early), Blask (C2, FAO 240-250; medium-early), Tur (C3, FAO 240; medium-early), Kozak (C4, FAO 250; medium-early), Bielik (C5, FAO 240; medium-early), Smok (C6, FAO 230-240; medium-early), SMH 220 (C7, FAO 230-240; medium-early) and Kresowiak (C8, FAO 240; medium-early), derived from free pollination and controlled self-pollination of maize. Tested cultivars come from two types of cross-breeding. One group of two way cross-breeds (SC) comprises the following: Bosman, Kozak, Smok and SMH 220. The other cultivars belong to the three way cross-breed group (TC). The experiment was established in the split-plot design in one replication, which prevented analysis of variance in the split plot system. In the middle of each plot the hybrids were grown without any space isolation. Plots were established side by side, which facilitated free open pollination between plants without isolators. After ear formation in the other half of the plots according to the experimental design, paper isolators were introduced to protect the ears against foreign pollen, while tassels were covered with paper bags to collect the pollen. During the pollination period hand pollination was performed according to the trial design. Tassels were shaken to collect the pollen in bags, followed by removing the isolators from the ears and covering them with bags containing the pollen. All the tested hybrids were grown following recommendations of good agricultural practice for maize crop. After full maturity of the plants ears were harvested, threshed and samples were prepared for laboratory analyses. Determination of physical traits In addition, 1,000 grain weight (TGW) and the hectoliter weight of all grain samples were recorded. TGW The effect of controlled self-pollination and open pollination on maize grain quality 493

increase between 1939 and 2004 was approximately 100 kg ha -1 · yr -1 , i.e. 2% per year (Tollenar & Lee, 2006). Currently maize hybrids up to the so called FAO 300, grown in Polish agriculture, come from leading breeders, both domestic and from the EU and the USA.
In experiments on the assessment of individual hybrids, maize plants are sown close together. Maize plants as cross and wind pollinated species can be pollinated with pollen from considerable distances, according to Felsot (2002) from up to 300 meters. However, experiments with coexistence of conventional and genetically modified maize showed that the main pollination is done within a much shorter distance, up to 20 m (Brookes et al., 2004;Arritt et al., 2007).
Grain endosperm is triploid (3n), of which 2n come from the male (transferred by pollen) and only 1n from the female plant, so a major impact of the male form can be expected on grain quality parameters. A good example of this relationship is the phenomenon of xenia observed on sweet maize pollinated by fodder maize. The xenia effect in endosperm and embryo traits has been known for a long time (Davarynejad et al., 1994;Bulant et al., 2000;David et al., 2001).
Maize grain is harvested from the F2 generation of the grown hybrid, but only if it is pollinated with the pollen of the same cultivar, as happens in large fields. For small plots used in variety trials, free open pollination by pollen from varieties sown in close proximity is observed and the resulting kernels are hybrids from the neighbouring varieties. In the case of twoway hybrids grown, random four-way hybrids are obtained, which may show some effects of heterosis as well as the effect of splitting of the traits. Thus, grain from such pollination may differ in its chemical composition and other characteristics from that obtained from self-pollination of only one variety. The results of this evaluation will not be comparable with those collected from large f ields. As it was observed by Mercer et al. (2002), there is a preference of other plant pollen (allopollen) over own pollen (autopollen) during fertilization of maize plants, with an average rate of 73.27% in the ear, which is close to the average frequency of 80.06% observed also by Balestre et al. (2007a).
The aim of this study was to determine the effect of pollen on grain quality. The research hypothesis assumes that maize grain produced from free pollination with pollen of different varieties grown next to each other on side-by-side plots varies in its chemical com-position and other characteristics from that produced by controlled self-pollination.  C7, and Kresowiak (C8, FAO 240; medium-early), derived from free pollination and controlled self-pollination of maize. Tested cultivars come from two types of cross-breeding. One group of two way cross-breeds (SC) comprises the following: Bosman, Kozak, Smok and SMH 220. The other cultivars belong to the three way cross-breed group (TC).

Plant material and experimental design
The experiment was established in the split-plot design in one replication, which prevented analysis of variance in the split plot system. In the middle of each plot the hybrids were grown without any space isolation. Plots were established side by side, which facilitated free open pollination between plants without isolators. After ear formation in the other half of the plots according to the experimental design, paper isolators were introduced to protect the ears against foreign pollen, while tassels were covered with paper bags to collect the pollen. During the pollination period hand pollination was performed according to the trial design. Tassels were shaken to collect the pollen in bags, followed by removing the isolators from the ears and covering them with bags containing the pollen. All the tested hybrids were grown following recommendations of good agricultural practice for maize crop. After full maturity of the plants ears were harvested, threshed and samples were prepared for laboratory analyses.

Determination of physical traits
In addition, 1,000 grain weight (TGW) and the hectoliter weight of all grain samples were recorded. TGW was determined using a Kopciuszek seed counter by Sadkiewicz Instruments (Poland). Three samples for each treatment were weighed on a Sartorius balance (by Sartorius Poland Company) accurate to 0.01 g. The hectoliter weight was assessed three times, using an electronic moisture and grain density test weight (GMDM) meter by Dramiński Electronics in Agriculture from Olsztyn, Poland.

Quality attributes
Contents of nutrients and starch in the grain were analysed at the laboratory of the Department of Agronomy, Poznaƒ University of Life Sciences (Poland). Starch content was determined by the polarimetric method according to Ewers (Gugal / a, 2002) using 0.31 N HCl and 4% phosphotungstic acid to precipitate protein substances. Sample analysis was performed in triplicate, while the blank test was performed once. Starch (S) content in dry matter (DM) percentage was calculated using the following formula: where P: angle of rotation of plane polarised light in the sample tested; P 1 : angle of rotation of plane polarised light in blank sample; α d : specific value of starch rotation (184.6); l: polarimeter tube length (dm); m: test sample weight (g); W: moisture content of sample (%).
Contents of organic components in the grain were analysed according to the following methods: total protein was determined by examining nitrogen content in the sample by the Kjeldahl method (Persson, 2008) and multiplied by a conversion factor of 6.25; crude fat with the Soxhlet method (Virot et al., 2008); crude fiber by hydrolysis of plant material residue; ash by dry incineration; and N-free extract (NFE) by subtracting from 100% the sum of the contents of the other components.

Statistical analysis
In the preliminary analysis of results recorded in laboratory assays mean values were calculated for parameters determining grain quality for the controlled and free open pollination variants. For the analysed traits the characteristics of variation were also determined, such as coefficient of variation, and minimum and maximum values.
The t-Student test for dependent samples was applied to compare pairs of means for grain quality parameters between the controlled and free open pollination combinations. At this stage of research this model of the statistical analysis has been used because the aim was to identify differences in the quality of grain depending on pollination type. The test for bound variables was selected, which reduced the effect of variation between genotypes and experimental units on the experimental error. Linear regression was also used to illustrate the relationship between treatments from the controlled and open pollination variants. Multivariate analysis was also used to demonstrate the combined effects of the parameters on the variation of the experimental treatments. Treatments refer here to combinations of levels of factors, where one factor is the cultivar and the other is the pollination type. Each of the treatments is described by eight traits, parameters characterising grain quality.
The hierarchical clustering method was applied, which allows to group similar objects, characterised by many traits. In this method object means were used for tested traits. Clustering was performed by agglomeration with the linkage method according to Ward. This method consists in linking objects in a manner minimising variability in the resulting group (Marek, 1989).
The applied Principal Component Analysis (PCA) makes it possible to reduce the number of traits (dimensions) describing the objects and illustrates the distribution of object means on the plane. The principal components facilitate a more synthetic assessment of the phenomenon, replacing a group of interrelated traits with one component, which may often be interpretable (Morrison, 1990). Statistical analysis was performed using the STATISTICA 10 software package. Table 1 presents mean values for the studied traits under controlled and free open pollination, their variability characteristics and the probability of rejecting a true hypothesis of equal means for the t-Student statistics. This test at a significance level of 0.01 showed significant differences between the means for the characteristics of the groups with controlled and free open The assumption of homogeneity of variance in the compared groups was also verified while none of the tested hypotheses about the equality of variance was rejected. Grain harvested from controlled self-pollination was characterised by a higher TGW and grain weight per hectoliter than that from the open pollination variant (Table 1). On average, the difference in the TGW between the self-pollination and open pollination combinations was 63 g and for the hectoliter weight it amounted to 1.8 kg hL -1 . Both differences were confirmed statistically. In addition, the use of self-pollination in this study resulted in the advantageous lower variability of TGW and hectoliter weight, at higher minimum and maximum values of the two traits in comparison with the open pollination variant.

Results
In our study pollination type determined starch content in grain of all maize hybrids compared (Fig. 1). The difference in starch content ranged from 1.7% The method of pollination determined total protein content in maize grain. Grain produced by controlled Self-polinated Open-polinated self-pollination contained 2.5% more protein than open pollinated maize and this difference was statistically significant. This relationship was found for all tested maize hybrids. Furthermore, protein content in DM of grain from self-pollinated hybrids was characterised by lesser variability and both higher minimum and maximum values in comparison with the grain harvested from openly pollinated plants.
The content of NFE, similarly to starch content of grain produced by open pollination, was significantly higher than in the self-pollinated hybrids, with the difference being on average 3%. Furthermore, in the case of open pollinated plants, this character showed more variability with a favourable increase in its minimum and maximum values.
Similarly, fat content in grain from the controlled self-pollination variant was on average higher by 0.3% than in grain produced by open pollination. However, this difference was not confirmed statistically (Table 1).
Also fiber content in DM of grain produced in the self-pollination combination showed only a trend towards higher values of these parameters as compared to grain produced by open pollination and these differences amounted to 0.2%. No differences were observed in ash content in DM of grain from either in the self-or open pollination variants.

Multivariate analysis
Sixteen medium-sized objects described by the eight studied traits were grouped using the clustering method and the methods of agglomeration at Ward branches. The result of the agglomeration is shown in a dendrogram (Fig. 2). It shows a clear distinction between the two experimental object groups. At the ends of the two major branches there are objects numbered C1 to C8 at the first branch (controlled pollination) and C11 and at the second branch there are objects from C9 to C16 (open pollination). All the cultivars in this analysis, except for no. C11, are grouped depending on the pollination type, which shows the extent to which this factor determines the value of produced grain.
Similar findings were obtained using PCA for medium-sized objects and 8 traits, in this study identified with their dimensions. 100% of variation between treatments is represented by eight principal components. The two first components explain 72.8% of variation and have values higher than 1 (Suppl. Fig. S1 [pdf online]), which allows us to reduce the number of traits of the objects and to illustrate the distribution of medium-sized objects on the plane.
The first principal component is composed in equal proportions of the following traits: NFE content in % of grain DM, TGW, protein % in DM and starch content in % of DM. The largest share in the second principal component (factor 2) is ash % of DM and fat % in DM, as shown in Table 2. We can say that our treatments were first of all differentiated by the traits that make up the first principal component. Fig. 3 shows how the object means are placed after projection of the first two principal components.   be seen that the objects were split into two groups, those with the letter k (average for the controlled pollination cultivars) are arranged on the left side of the vertical axis and the other group of objects is on the right (except for the open pollinated cv. Tur). This confirms the conclusion provided by the clustering method.

Discussion
Maize germ and especially endosperm, which is more than 95% of kernel mass, presents the phenomenon of xenia, which is a direct expression of pollination. The xenia effect can be interpreted as an early expression of heterosis, which increases the ability of the endosperm, modif ied by the effect of foreign pollen. Cross pollination directly modifies the accumulation of photoassimilates and thus affects kernel weight. The greater the genetic difference between the allelopollen receiver and the donor plant, the greater the chances are that the phenomenon will occur (Denney, 1992).
This increase may result from the activity of the enzyme ADPGPPase (EC2.7.7.27) generally responsible for the synthesis of starch in the endosperm of maize kernels (Bulant et al., 2000). According to those authors, ADPGPPase activity in grain from foreign pollination was higher, which resulted in increased grain weight. The activity of this enzyme in kernels produced by foreign pollination (allelopollen) was by 19% higher 14 days after pollination, and the difference decreased to 8% after 74 days, compared with the kernels formed as a result of self-pollination.
A study by Balestre et al. (2007b) showed that foreign pollen (allopollen) caused the average kernel weight to increase by 16.25 mg when compared to selfpollination, which was statistically confirmed at 95% probability. According to the authors, this increase is caused by several factors. One of them may be inbreeding depression, which occurs when the plant is pollinated with its own pollen. In other words, lethal genes (deleterious alleles), whose expression is masked by heterozygous loci in hybrids, have a chance to be revealed at self-pollination. This is likely to occur especially in single cross hybrids, which have the strongest effect of heterosis in the F1 generation, and plants are assumed to have the same genotype. In such a situation on the field (cultivation of SC hybrids), even with cross pollination, plants do not differ from selfpollinated plants (the same genotype). The combinations of dominant or partly dominant alleles can also affect specific phases or grain development processes that reduce the final weight (Leng, 1949).
In a study of Balestre et al. (2007a) the heterosis effect caused by foreign pollination (allopollen) was 4.65%, compared to self-pollination (autopollen). This result was lower than in another study of Balestre et al. (2007b), where it amounted to 11.90%.
The results of our studies carried out on Polish hybrid varieties did not confirm the results mentioned above, as TGW resulting from self-pollination was higher by an average of 63 g compared with those produced by open-pollination. Similar results were obtai-The effect of controlled self-pollination and open pollination on maize grain quality 497 ned by Tsai & Tsai (1990), Weiland (1992) and Andrade & Pereira (2005), who conducted research on single, double and triple cross hybrids, and also did not observe the impact of allopollen on reciprocal crosses. The probable reason, which explains the different response of hybrids to pollination method observed by researchers, i.e. an interaction with the heterosis effect in the embryo and endosperm, is connected with a hybrid or line combining ability. It is known that hybrids and lines differ in their general and specific combining ability. In other words, there are differences in heterosis, which are associated with dominance and divergence in lines and then in hybrids derived from them. Thus, in the most contrasting pairs an increase in grain weight is observed. Changes in grain weight depend on the specific combining ability of these hybrids. The effect of changing kernel weight may vary, depending on the crossing involved between the hybrids (Hoekstra et al., 1985;Pinter et al., 1987). The effect of self-pollination may be different, since the proportion between the maternal and paternal genomes differs in the triploid endosperm.
Significant differences in grain weight, their size and test weight between maize hybrids depending on the method of pollination were also reported in a study by Letchworth & Lambert (1998). Test weight (average for 12 hybrids) was higher in the case of self-pollination compared with kernels produced by open pollination. The authors cited the result obtained for ten hybrids; in two cases the result was opposite. Similar relations as for test weight were reported by those researchers for kernel weight and kernel size, which corresponds with our own results, in which both of these traits have higher values in the case of selfpollination.
Based on the observations of an increase in grain weight as a result of open-pollination, studies were undertaken to obtain increased grain yield on production f ields by sowing mixtures of cultivars (Bulant & Gallais, 1998;Mercer et al., 2002;Balestre et al., 2007b). In a study by Andrade & Pereira (2005) the effect of changes in kernel weight as a result of foreign pollination did not occur. While in the study with other species, researchers observed that berries of blueberry were larger after cross-pollination than after selfpollination. All fertility parameters were reduced for self-pollination, including a lower number of seeds per berry that contributed to a smaller berry size (Chavez & Lyrene, 2009). Signif icant differences in kernel weight of Prunus dulcis, depending on the method of pollination were also reported in a study by Dicenta et al. (2002). These authors observed obvious differences between pollination treatments, kernels from crosspollination being heavier (mean weight 1.48 g) than those from self-pollination (1.37 g).
However in a study by Derin & Eti (2001) the effect of changes in pomegranate fruit quality as a result of foreign pollination did not occur. Different sources of pollen did not play an important role in the quality parameters, namely, juice percentage, total soluble solids (TSS) and titrable acidity contents. On the other hand, fruit weight and 100 aril weight were increased considerably by cross-pollination with pollen taken from other cultivars' male flowers.
The method of pollination differentiated chemical composition of maize grain. Maize hybrids differ in protein and starch contents in the grain, and the dependence of the concentration of these components in the grain is inverse. Similarly, an inverse relation is observed between protein content and grain yield (Idikut et al., 2009). In a study by Letchworth & Lambert (1998) grain produced by self-pollination contained more protein than the open-pollinated variant and the ranking of hybrids in terms of protein content did not change.
The evaluation of reverse crossing shows a strong maternal effect on protein content and a lack of the pollen effect. In our studies, as in the cited literature, grain produced by controlled self-pollination contained more protein than the DM produced by a mixture of pollen of the variety and other varieties growing nearby.
Also fat content in grain from controlled self-pollinated plants was higher than in the case of open pollination. This result does not correspond to that reported by Letchworth & Lambert (1998), who showed significantly higher fat contents in grain produced by open pollination than self-pollination. In the study of Letchworth & Lambert (1998), grain produced by open pollination had higher starch content than that from self-pollinated plants. Reverse crosses showed the maternal effect on starch content and the absence of the effect of pollen. Usually the result of allopollen in reciprocal crosses is not significant and inversion of the female plant does not affect kernels (Tsai & Tsai, 1990;Weiland, 1992;Andrade & Pereira, 2005).
Open pollination by a pollen mixture from varieties grown nearby led, similarly as in the cited literature, to formation of greater amounts of starch in the kernels than in controlled self-pollination. Also NFE content in the grain produced by open pollination was significantly higher than in the case of self-pollination. Our studies showed that contents of fat, fibre and ash in the grain produced by open pollination with a mixture of pollen from varieties grown in close proximity were slightly lower than in kernels derived from self-pollination. These differences, however, were not confirmed statistically. Opposite observations were obtained by Letchworth & Lambert (1998), who showed a higher fat content in the kernels resulting from openpollinated than from self-pollinated plants. Significant differences between the hybrids evaluated in the cited studies indicate the maternal effect, and the data obtained for reciprocal crosses indicated the pollen effect. Fat content in the grain of tested hybrids varied within a narrow range of 41-55 g kg -1 and the ranking of the three hybrids with the highest and lowest fat contents did not change depending on the method of pollination. According to these authors, detailed studies on fat content should be conducted in self-pollination combinations. Analyzing other species in the literature we can also find data that sunflowers pollination by bees improves seed weight and seed oil content. The ability of honey bees to forage on many sunflower heads improves their pollination potential. This study showed that bee pollination increased sunflower seed number by 59% (Nderitu et al., 2008). Also interesting results were obtained by Oukabli et al. (2002) on Prunus dulcis over a two years study. The average fruit set after cross-pollination (25.6%) was higher than that obtained by self-pollination (19.2%). The low fruit set after self-pollination suggests the existence of inbreeding effects.
According to David et al. (2001), maize hybrids grown in the vicinity of other hybrids, produce higher quality grain yield than in the case of self-pollination. Our own studies confirmed that open pollination was a factor increasing starch concentration in the grain.
The results of this study and literature data indicate a need to clarify the methodology of variety testing, especially in the allocation of the experimental varieties on plots. When it is difficult to meet the isolation requirement, the solution might be to use isolators and forced self-pollination to produce grain, which will be subjected to chemical analyses. Kernels obtained in this way will reflect the quality of the grain specific to the tested hybrid. In addition, these results will correspond with the results obtained in farming practice when hybrids are grown in large fields.
In conclusion, the study confirmed differences in the quality of grain of maize hybrids obtained by selfpollination or open pollination. Kernels obtained by open-pollination were characterised by higher contents of N-free extract and starch, and lower protein contents. To ensure that the results of variety testing are in line with field production conditions, it is necessary to verify methodology of maize variety testing. Further such studies may indicate specific recommendations for agricultural practice, such as mixtures consisting of selected hybrids with good combining abilities, which will contribute to improved grain quality with no additional costs incurred.