The base population was the multicolor maize synthetic population EPS4. This synthetic was made in 1981 by mixing 100 kernels from each of three open-pollinated populations (Salcedo, Taboadelo and Cambados) from northwestern Spain with diverse kernel colors, followed by random mating for more than 10 generations ( Rodríguez et al. , 2013 ). Two random 300-kernel samples of white and black kernels were separated from the population EPS4. These two samples were multiplied independently by plant-to-plant pollinations, using each plant solely as either male or female. The separated subsamples of white and black kernels were cleaned and grinded in a laboratory grinding mill. The flour was screened with a sieve of one millimeter in diameter.

Black kernels had high anthocyanin content 5.38±0.018 μmol/g fresh weight while white kernels had only 0.28±0.018 μmol/g, while the difference in carotenoid content was lower (1.76±0.064 vs 0.75±0.064 μmol/g in black and white kernels, respectively) ( Rodríguez et al. , 2013 ). Differences in pigment content among kernels with diverse colors explained most differences in AA, as the AA of the hydrophilic fraction was highly correlated to that of the pigments, while differences among kernel colors for the AA of the lipophilic fraction were not significantly different. Concerning the nutritive composition, this maize variety had around 11% of total protein, 5.5% of total fat, 1% of brute fiber, and 58% starch, without significant differences among kernel colors.

Two laboratory bioassays were conducted to investigate the effect of AA of white and black maize flour on growth and development of S. nonagrioides larvae. We used the artificial diet published by Farinós et al. (2004) as control treatment. The artificial diet was made starting with 1 liter of hot water, and then we added 2.5 g of benzoic acid and, after reaching the boiling point, 26 g of agar while homogenizing the mixture with an electric mixer. We introduced the bowl in cold water and, while the mix was cooling, we added 43 g of yeast, 40 g of wheat germ and 160 g of maize flour while homogenizing with the electric mixer. The control treatment was made with a type of standard commercial maize flour called polenta. While the mix continued cooling, we added 6 g of ascorbic acid, 1 g of nipagin (from Sigma) and 1.55 g of Wesson salt mixture (from Sigma). When the mix reached 50 °C and was homogeneous, we poured it into plastic boxes.

In the first bioassay, for each maize flour treatment, the polenta of the artificial diet was replaced with the same amount of white or black maize flour, respectively. A second sample of each (white or black, respectively) maize-diets included vitamin C (10 g/L). Therefore, there were five different diets: control, white maize, white maize + vitamin C, black maize, and black maize + vitamin C. The experiment followed a randomized complete block design with four repetitions. Sixty neonate larvae per repetition and treatment were randomly assigned to each diet by placing a 1 cm-diameter piece of diet plus one larva in each individual 35 mm-diameter Petri dish. The neonate larvae were obtained by taking eggs from maize plants grown in a greenhouse with a large population of moths hatched from larvae collected from the field; eggs were placed on pieces of artificial diet made following the recipe shown above for the control diet; when the larvae were hatched are reached a 2-mm size, the alive larvae were taken for the bioassay. The piece of diet was changed by a fresh piece two times per week in order to allow the larvae to eat fresh diet ad libitum . After two weeks, half of the larvae (30) of each treatment continued to be fed on the same diet while the other larvae were fed with the same diet but supplemented with 5% H ₂ O ₂ in order to produce oxidative stress; this concentration of H ₂ O ₂ is enough for inducing oxidative stress ( Rao et al. , 1997 ). Initial larval weight was recorded and then the weight of each larva was taken two times per week. At the same time we took larvae weights, we changed the piece of diet during four weeks, renewing the piece of diet and recording the number of dead larvae or pupa; therefore we took up to 8 weights per larvae.

A second bioassay was carried out in order to double check the effects of white vs . black maize flour, avoiding the interference of any other factor. In this bioassay we followed a simplified version of the first bioassay, wherein only the white and the black-flour treatments were performed. The recipes and methods were as in the first bioassay. This second bioassay was analyzed separately because the experimental design was different and, therefore, they could not be combined in a single analysis. Furthermore, the second bioassay intended to simplify the conditions in order to allow a direct comparison between black and white flour without interactions from other compounds.

Antioxidant activity of the diets was tested on random samples of each of the five diets with and without H ₂ O ₂ . For doing so, 5 g of each fresh diet were lyophilized. As the bioassays ended, lyophilized diets were grinded in an analytical grinding mill (Model A10, IKA, Germany). Two subsamples were taken from each sample and antioxidant capacity analyses carried out per duplicate. Freeze-dried and ground samples (10 mg) were extracted with 1 mL of 80% aqueous methanol in dark maceration for 24 h. After centrifugation (3700 rpm, 5 min), methanolic extracts were employed in order to determine AA by using four methods: ferric reducing antioxidant power (FRAP) ( Benzie & Strain, 1996 ); 2,2-azinobis (3-ethyl-benzothiazoline-6-sulfonic acid) (ABTS) ( Miller & Rice-Evans, 1997 ); 2,2-diphenyl-1-picrylhydrazyl (DPPH) ( Brand-Williams et al. , 1995 ); and Folin-Ciocalteu (FOLIN) ( Singleton & Rossi, 1965 ). We used the four methods of AA determination following Sotelo et al. (2014) because, although normally they are highly correlated, these four methods measure slightly different aspects of AA, for example, with DPPH method, the reaction takes place in methanol, where for the other three methods, reaction occurs in water. Nevertheless, FOLIN is the most common method for AA determination.

All AA assays were carried out in a microplate spectrophotometer (Spectra MR; Dynex Technologies, Chantilly, VA, USA). Standards prepared with different concentrations of Trolox (0, 0.008, 0.016, 0.024, 0.032 and 0.04 mM) were measured for FRAP, DPPH and ABTS analyses and AA values were normalized to Trolox equivalents per gram of dry weight. Standards prepared with different concentrations of gallic acid (0, 0.008, 0.016, 0.024, 0.032 and 0.04 mM) were also measured. Results of FOLIN assay were expressed in terms of micromoles of gallic acid equivalents per gram of dry weight.

Repeated measures analysis was used to analyze the weekly weight measurements of larvae. A growth curve of weight on time was estimated for each treatment, and homogeneity of linear and quadratic coefficients was tested for each pair of treatments. The analysis was made using the MIXED procedure of SAS (2008) . All factors were considered random except treatment which was considered fixed. The covariance was calculated following Littell et al. (1996) . Additionally, analysis of variance were made for each bioassay by using the using the MIXED procedure of SAS (2008) , considering repetitions as random effects and color grains as fixed effects. And comparisons of means among treatments were calculated for larval weight for each time with the Fisher's protected LSD.

Finally, to analyze larval survival, the Kaplan–Meier estimates of the survival function were calculated for each treatment, and curves were compared using the log rank test ( Cantor, 1997 ; Ordás et al. , 2002 ). Survival functions were significantly different if they deviated from the expected values of the null hypothesis (that survival functions are equivalent in all treatments). The statistic determines whether differences between survival functions are significantly different at any probability level, thus indicating that larvae have significantly different survival in some treatments than in others (LIFETEST procedure of SAS). Missing larvae were censured for larval survival analysis, which means that the analysis considered that larvae lived at least until they disappeared. When larvae reached pupal stage, the larvae survival was scored as reaching 30 days.

The four measurements of AA were strongly correlated, with highly significant correlations coefficients that varied from r ² = 0.832 between FOLIN and DPPH, to r ² = 0.996 between FRAP and ABTS. Correlations were above 0.9 for ABTS and DPPH (r ² = 0.979, p < 0.001), DPPH and FRAP (r ² = 0.979, p < 0.001), ABTS and FOLIN (r ² = 0.928, p < 0.001), and FRAP and FOLIN (r ² = 0.920, p < 0.001). These values are in agreement with previous results indicating that correlations between methods for measuring AA were very high ( Huang et al. , 2005 ; Kusznierewicz et al. , 2008 ; Soengas et al. , 2012 ). Our coefficient correlations were higher than those previously reported for other crops ( Zhi-Xiang et al. , 2011 ; Sotelo et al. , 2014 ), indicating that these measurements were highly consistent and reliable. Since the four measures of AA were highly correlated, we will focus on FOLIN (Table 1, Fig. 1).

Our results show that polenta had the highest AA, and that vitamin C had stronger AA than polyphenols. The recipe for the control diet already includes vitamin C because it has been designed for optimum growth of larvae (Farinós et al. , 2004). Adding H ₂ O ₂ to the diet resulted in a non-significant reduction of AA, even though we used higher concentration of H ₂ O ₂ than previous reports ( Rao et al. , 1997 ). Vitamin C always increased AA, significantly for white flours. White flour had lower AA than black flour but, when an external antioxidant, such as vitamin C, or when an external oxidant, such as H ₂ O ₂ , is added, the order of AA is reversed in black and white flour because AA of vitamin C hides that of polyphenols. This could be due to an interaction between two antioxidants such as vitamin C and polyphenols in black flours or that the processing of the maize flour alters the AA of the polyphenols. Accordingly, several authors have shown that a cooking process, such as nixtamalization, reduces pigment content and antioxidant capacity ( Del Pozo-Insfran et al. , 2006 ; De la Parra et al. , 2007 ; López-Martínez et al. , 2011 ; 2012 ). The causes of AA loss can be temperature and acidity, which have been reported as agent of antioxidant reduction by Li et al. (2011) who reported that citric acid significantly affects polyphenols content and AA. Furthermore, adding maize flour to the diet implies a dilution of AA that minimizes the presumable AA of polyphenols, as previously shown by Rodríguez et al. (2013) , who reported a reduction in pigments' content and their respective antioxidant capacity due to dilution. Furthermore, as Petroni et al. (2014) stated, one major limitation in assigning a health property to polyphenols is the influence of other metabolites in the diet, acting as possible confounding factors. Indeed, there could be interactions between antioxidant substances, as indicated by Huang et al. (2005) .

In the first bioassay, the variation of larval weight as they feed on diets, indicate that the larvae always grow more in the control diet, designed for optimum nutrition of larvae, than in the maize-diets, and the addition of H ₂ O ₂ has not significant effects on larval growth (Table 2). Besides, larvae fed on flour without oxidative stress (without H ₂ O ₂ ) have grown more in diets with more antioxidants (polyphenols and / or vitamin C). The difference in growth coefficient between black flour + vitamin C and white flour was significant. In contrast, when larvae were fed on diets with oxidative stress (with H ₂ O ₂ ), the larvae grew more in the diet with white flour + vitamin C. We would not expect that this diet have had the highest antioxidant concentration but it was the diet with H ₂ O ₂ with the highest AA. This is probably due to vitamin C - polyphenols interaction discussed above.

In the second bioassay, differences were neither significant between coefficients of regression of larval weight nor between larval weight at any stage of development, corroborating the previous results which showed that differences among black and white flour alone were not uncovered with this experimental design.

Larval weight was not related to survival as shown by the low correlation between larval weight and Log-rank (r ² = 0.30, p =0.392). Survival analyses indicated that the proportion of dead larvae was lowest for the control diet with or without H ₂ O ₂ (Table 2). The values of Log-rank were also negative for white maize with or without H ₂ O ₂ and for black maize without H ₂ O ₂ . Mortality was highest for both white and black maize irrespective of the addition of vitamin C or H ₂ O ₂ . These results show that the relationship between AA and mortality was not clear; in fact, rank correlations between AA and Log-rank was not significantly different from zero (r ² = -0.14, p =0.701).

The weights recorded in the first bioassay showed that the larvae grown in the control diet had always the highest weights (Table 3). At first stages of development, higher concentration of antioxidants may be detrimental to larvae, i.e . larvae fed on white flour weighed significantly more than those fed on black flour with vitamin C. As larvae grew up, black flour with vitamin C increased larval weight to values that were not significantly different from the control from week sixth to eighth, and so did white flour with vitamin C for weeks seventh and eighth. Therefore, vitamin C has positive effects on larval growth. Besides, there was a weak relationship between AA and larval weight; actually, rank correlations between AA and larval weight were positive and moderate (r ² = 0.51, p =0.136).

The effects of AA of polyphenols on growth of corn borers can be due to a possible insecticide effect as proposed by Bedini et al. (2016) or to the positive effect of antioxidant nutrients on health as shown by Petroni et al. (2014) . As previous authors have shown, polyphenols could have both positive and negative effects on larval health and growth, depending on other substances that might interact with them ( Costa-Arbulú et al. , 2001 ; Karageorgou et al. , 2008 ; Lev-Yadun & Kevin, 2008 ).

As conclusion, these results indicate that during the first stages of larval development, antioxidants have an insecticide effect, while, as the larvae grow up, antioxidants favors larvae growth. Furthermore, the antioxidant activity of maize polyphenols interacts with other antioxidant substances.