rbz Revista Brasileira de Zootecnia R. Bras. Zootec. 1516-3598 1806-9290 Sociedade Brasileira de Zootecnia 01813 10.37496/rbz5520250253 Forage crops Variability and prediction of energy nutritional traits in grains from maize genetic basis and their relationships with meteorological variables 0000-0003-0241-4226 Loro Murilo Vieira Conceptualization Data curation Investigation Software Validation Writing – original draft 1 * 0000-0002-8608-9960 Cargnelutti Alberto Filho Conceptualization Investigation Software Validation Writing – review & editing 1 Universidade Federal de Santa Maria Departamento de Fitotecnia Santa Maria RS Brasil Universidade Federal de Santa Maria, Departamento de Fitotecnia, Santa Maria, RS, Brasil. muriloloro@gmail.com

Gustavo José Braga

Leonardo Simões de Barros Moreno

The authors declare no conflict of interest.

 13 11 2026 2026 55 e20250253 22 12 2025 10 06 2026 Copyright: This is an open access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. ABSTRACT

The purpose of this study was to evaluate whether there is variability in the energy and nutritional composition of maize grains among different genetic backgrounds and sowing dates and to determine whether there are relationships between energy nutritional traits and meteorological variables. In total, 773 grain samples from four maize genetic backgrounds, cultivated across ten sowing dates in the years 2021, 2022, and 2023, were evaluated. Descriptive statistics were calculated, and the principal component, correlation, multiple linear regression, and random forest analyses were performed. The contents of ash, ether extract, crude fiber, starch, amylose, and apparent metabolizable energy corrected for nitrogen balance varied among maize genetic backgrounds and sowing dates. At the study site, maize sown in January and February had the highest levels of starch, amylose, and apparent metabolizable energy corrected for nitrogen balance, and the lowest levels of ash, ether extract, and crude fiber. Single-, triple-, and double-cross hybrids had the highest contents of starch, amylose, and apparent metabolizable energy corrected for nitrogen balance, whereas open-pollinated varieties had the highest contents of ash, ether extract, and crude fiber. The contents of crude fiber, starch, amylose, and apparent metabolizable energy corrected for nitrogen balance in maize grains could be predicted using meteorological variables across all maize genetic backgrounds. Random forest models exhibited high predictive accuracy (R2 ≥ 0.78), with precipitation between female flowering and harvest emerging as the main meteorological variable associated with the contents of crude fiber, starch, amylose, and apparent metabolizable energy corrected for nitrogen balance (IncMSE ≥ 23.83%).

 Keywords energy nutritional composition genetic variability plant breeding Zea mays CNPq 304652/2017-2 CNPq 304878/2022-7 CAPES 001 CAPES CAPES/PROEX/88881.009488/2024-01 Adisseo Brasil Nutrição Animal company CEAN Laboratory To the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq - Processes 304652/2017-2 and 304878/2022-7) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) - Financing Code 001; CAPES/PROEX/88881.009488/2024-01. To the scholarship-holding students and volunteers for their help in collecting data. To the Adisseo Brasil Nutrição Animal company, especially to the CEAN Laboratory for carrying out the laboratory analyses. 1. Introduction

Maize (Zea mays L.) is the main energy source in rations for monogastric and ruminant animals. The contents of ash, ether extract, crude fiber, starch, amylose, and apparent metabolizable energy corrected for nitrogen balance in maize grains determine their nutritional value and influence feed efficiency in animals (Mehta et al., 2020; Guo et al., 2023). However, the nutritional composition of maize grains results from the interaction between genetic and environmental factors, which modulate grain quality and, consequently, animal performance. Maize genetic backgrounds, such as single-, triple-, and double hybrids, and open-pollinated varieties (populations maintained by natural pollination) influence the nutritional composition of the grains. For example, single and triple hybrids may have higher ash contents, with single hybrids tending to have higher ether extract contents and lower carbohydrate contents, while triple hybrids show the opposite pattern (Gonçalves et al., 2003). This compositional variation among genetic backgrounds highlights the maize genetic diversity potential of maize genetic diversity for the development of genotypes adapted to different production contexts, considering nutritional, industrial, and environmental demands.

Meteorological conditions, especially during the reproductive stage, affect the energy content of maize grains. Variation in water availability among locations and sowing dates determines the physiological and metabolic processes involved in starch deposition and final grain quality (Martínez et al., 2017; Álvarez-Iglesias et al., 2021; Safian et al., 2022; Guo et al., 2023). Accumulated rainfall during the reproductive phase is one of the main factors that determine starch deposition, as water deficits compromise the activity of biosynthetic enzymes, reduce carbohydrate formation, and increase crude fiber content, thereby reducing digestibility (Yang et al., 2003; Chaves et al., 2009; Jeon et al., 2010). Despite these advances, studies that evaluate, in an integrated manner, the effects of genetic basis and meteorological variables on the energy composition of grains are still scarce, limiting the strategic use of this information in crop breeding and management.

In this context, understanding these complex interactions remains a major challenge, particularly due to the multivariate nature of the relationships between environmental conditions and plant performance. Recent advances in statistics and data science have enabled more precise approaches investigating these relationships. Models such as the stepwise multiple regression, and machine learning algorithms, such as random forest, have been successfully applied to model and predict agronomic and nutritional traits using environmental data (Maseko et al., 2024; Sun et al., 2025; Zelaya Arce et al., 2025). These tools help identify relevant environmental variables, estimate nutritional traits before harvest, and support field decision-making, such as the selection of appropriate sowing dates and genotypes for local conditions.

Understanding the variability in the nutritional composition of grains among genetic backgrounds and sowing dates, as well as identifying the meteorological factors that most influence this composition, is essential for the development of more efficient, economical, and sustainable production systems. This approach can also reduce laboratory testing costs, improve feed formulation, increase feed efficiency, and support crop adaptation to weather extremes. Therefore, our objectives were to evaluate whether there is variability in the energy nutritional composition of grains among maize genetic backgrounds (single hybrids, triple hybrids, double hybrids and open-pollinated varieties) and sowing dates, whether relationships exist between energy and nutritional traits and meteorological variables.

 2. Material and methods 2.1. Location, dates, and years of the experiments

Maize genotypes were evaluated on ten sowing dates: September 21, 2021, October 20, 2021, November 20, 2021, December 20, 2021, January 30, 2022, September 6, 2022, October 14, 2022, November 24, 2022, December 30, 2022, and February 6, 2023 (years 2021, 2022, and 2023), at the experimental field of the Department of Plant Science of the Universidade Federal de Santa Maria. Except the September 21, 2021 sowing date, when 71 genotypes were sown, 78 maize genotypes, including single-, triple-, and double-cross hybrids and open-pollinated varieties, were sown on each of the other dates.

On each sowing date, maize genotypes were sown in single-row plots. Each plot consisted of a 5-m long row spaced 0.80 m between rows and 0.20 m between plants in the row, totaling 25 plants per plot. In this study, the genotypes were replicated on ten sowing dates. The area was harrowed and fertilized with 415 kg ha−1 chemical fertilizer, with the formula (NPK) 05-20-20. Nitrogen fertilization with urea (N - 46%) was split into two applications: 250 kg ha−1 at the V4 stage and 150 kg ha−1 at the V6 stage.

 2.2. Assessment of the grain energy nutritional traits

At the harvest point, a 100-g sample of maize grains was collected from each genotype at each sowing date (773 samples), placed in a paper bag, and dried in a forced-air oven until reaching 13% moisture. Subsequently, the grains were ground in a Retsch Mill (model ZM 200) fitted with a 1-mm sieve. The ground samples were then scanned using a FOSS NIRS DS2500 instrument (Adisseo Brasil Nutrição Animal) by near-infrared reflectance spectroscopy (NIRS). Based on the absorbance spectra generated for each sample, nutrient composition was determined using multivariate calibration for ground maize through the Precision Nutrition Evaluation (PNE) platform. Thus, ash (Ash), ether extract (EE), crude fiber (CF), starch (ST), and amylose (AML) contents were determined for each sample and expressed as g/100 g, whereas apparent metabolizable energy corrected for nitrogen balance (AMEn) was expressed as kcal/kg. Detailed information on the wet chemistry analyses used to calibrate the NIRS models can be obtained from Oviedo-Rondón et al. (2024).

 2.3. Meteorological variables

The meteorological variables were obtained from the Instituto Nacional de Meteorologia (INMET) station, located 100 m from the experimental area. For each day, the hourly global solar radiation, in MJ m−2 hour−1, the hourly maximum and minimum temperatures, in °C, and the rainfall, in mm, were obtained for the period between sowing and the harvest point of the genotypes, corresponding to each sowing date. For each day, the daily global solar radiation, in MJ m−2 day−1, was calculated by summing the hourly global solar radiation, in MJ m−2 hour−1. Afterwards, for each genotype, on each sowing date, based on the daily global solar radiation, the accumulated global solar radiation was calculated for the following subperiods: Sowing to female flowering (RSFF, MJ m−2) and female flowering to harvest (RFFH, MJ m−2).

For each day, 24 hourly mean temperature values were calculated from the hourly maximum (Tmax) and minimum (Tmin) temperatures. Then, the mean daily air temperature (Tmed), in °C, was calculated by averaging these 24 hourly mean temperature values. Subsequently degree days (GD, in oC) were calculated according to the method proposed by Arnold (1960) using the following expression:

GD = Tmed - Tb,

in which Tmed = average daily air temperature, in oC and Tb = lower-base temperature (10 °C) . Tb is the temperature below which no plant growth occurs and if any growth occurs, it occurs at a significantly reduced rate. When Tmed ≤ Tb, GD was considered equal to zero. For each genotype, on each sowing date, the thermal sum was obtained by adding the GD values of each of the following subperiods: Sowing to female flowering (SSFF, °C day) and female flowering to harvest (SFFH, °C day). For each genotype, on each sowing date, the accumulated rainfall was calculated in each of the following subperiods: sowing to female flowering (PSFF, mm) and female flowering to harvest (PFFH, mm).

 2.4. Statistical analysis

For each energy nutritional trait, on each sowing date, the following statistics were calculated: minimum, mean, maximum, and coefficient of variation. Afterwards, for each trait, on each sowing date and genetic basis, the average was calculated. The trait means between sowing dates regardless of genetic basis, between genetic basis on each sowing date, and between genetic basis regardless of sowing dates were compared using the Scott-Knott test at 5% significance. The principal component analysis was applied to group crops by harvest, genetic basis, sowing months and sowing dates in relation to energy nutritional traits and meteorological variables. For each maize genetic basis, Pearson linear correlation networks were developed between pairs of traits.

For each genetic basis, parameters of the stepwise multiple linear regression model with backward selection were estimated, considering the traits Ash, EE, CF, ST, AML, and AMEn as dependent variables and the meteorological variables (RSFF, RFFH, SSFF, SFFH, PSFF, PFFH) as independent variables. For each genetic and general basis, the random forest algorithm was used to estimate the importance of variables RSFF, RFFH, SSFF, SFFH, PSFF, and PFFH for predicting energy nutritional traits. The importance of the variables was estimated based on the percentage increase in the mean square error (IncMSE). The model was fitted with 500 trees (ntree = 500), selecting three random variables per node (mtry = 3) and activating the option to calculate the variable importance (importance = TRUE).

For each multiple linear regression and random forest model, the coefficient of determination (R2), the root mean square error, the coefficient of determination of the leave-one-out cross-validation (R2cv), and the root mean square error of the leave-one-out cross-validation (RMSEcv) were calculated, which generates the model with n-1 observations and evaluates it using the remaining observation. This procedure was repeated n times, excluding a different observation at each time. All statistical analyses were performed using R software (R Core Team, 2025).

 3. Results 3.1. Variability of energy nutritional traits

The energy nutritional composition varied significantly between sowing dates (Table 1). The contents of ash (Ash), ether extract (EE), crude fiber (CF), starch (ST), amylose (AML), and apparent metabolizable energy corrected for nitrogen balance (AMEn) ranged between sowing dates, with overall mean of 1.121 g/100 g, 3.871 g/100 g, 2.438 g/100 g, 63.503 g/100 g, 31.582 g/100 g, and 3,292.62 kcal/kg, respectively. The content of each trait changed specifically depending on the environment, with emphasis on EE, whose highest value (4.029 g/100 g) was observed for the November 20, 2021 sowing date. However, this date did not provide maximum values for the other traits. Among the genotypes and sowing dates, grain yield ranged from 0.12 Mg ha−1 to 9.13 Mg ha−1, with an average of 3.63 Mg ha−1.

Minimum (Min), mean (Med), maximum (Max) contents and coefficient of variation (CV, %) of energy nutritional traits in grains of single, triple, and double cross hybrids and open-pollinated varieties sown on ten sowing dates in the years 2021, 2022, and 2023
Date Ash   EE   CF
Min Med Max CV Min Med Max CV Min Med Max CV
09/21/2021 0.890 1.114b 1.250 6.67   3.180 3.841c 4.710 5.97   1.870 2.256c 2.620 6.87
10/20/2021 0.980 1.158a 1.350 6.35   3.340 3.958b 4.590 5.99   1.860 2.337b 2.800 6.97
11/20/2021 0.940 1.129b 1.290 6.61   3.410 4.029a 4.830 6.78   1.900 2.297c 2.680 6.64
12/20/2021 0.860 1.074c 1.300 8.24   3.260 3.912b 4.540 7.43   1.950 2.278c 2.750 7.28
01/30/2022 0.920 1.107b 1.290 7.11   3.120 3.748c 4.260 7.26   1.920 2.240c 2.580 6.95
09/06/2022 0.890 1.100b 1.270 7.44   3.520 3.942b 4.710 5.91   2.250 2.633a 3.310 8.81
10/14/2022 0.960 1.171a 1.370 6.94   3.440 3.900b 4.460 5.56   2.330 2.666a 3.240 5.72
11/24/2022 0.950 1.166a 1.430 8.30   3.410 3.938b 4.620 5.65   2.310 2.663a 3.350 7.37
12/30/2022 0.940 1.112b 1.400 8.47   3.060 3.789c 4.550 7.14   2.170 2.636a 3.320 7.82
02/06/2023 0.910 1.077c 1.220 6.97   3.100 3.647d 4.750 7.82   1.870 2.352b 3.400 9.46
General 0.860 1.121 1.430 7.88   3.060 3.871 4.830 7.12   1.860 2.438 3.400 10.42
Date   ST       AML       AMEn  
Min Med Max CV Min Med Max CV Min Med Max CV
09/21/2021 62.310 64.664b 66.330 0.93   30.150 34.154a 37.140 4.64   3273.00 3358.620a 3406.00 0.77
10/20/2021 63.250 64.372c 65.520 0.78   28.290 31.422e 35.760 5.24   3291.00 3331.128b 3370.00 0.55
11/20/2021 63.690 64.618b 65.540 0.72   30.390 33.487b 36.410 3.76   3310.00 3352.538a 3384.00 0.61
12/20/2021 63.250 64.989a 66.180 1.00   29.160 32.943c 35.450 3.94   3264.00 3364.885a 3430.00 1.06
01/30/2022 62.600 64.311c 65.620 0.93   28.800 31.809e 35.320 4.09   3200.00 3359.141a 3422.00 1.49
09/06/2022 63.540 64.964a 66.830 1.24   27.180 31.437e 34.830 4.80   3294.00 3364.603a 3419.00 0.92
10/14/2022 60.110 61.299e 62.420 0.90   24.280 29.309g 34.090 6.24   3060.00 3176.051d 3267.00 1.35
11/24/2022 59.760 61.430e 63.460 1.06   24.470 28.913g 32.880 5.65   3059.00 3169.333d 3253.00 1.44
12/30/2022 58.960 61.365e 63.150 1.38   24.040 30.089f 34.210 5.67   3060.00 3183.385d 3269.00 1.47
02/06/2023 59.340 63.121d 66.250 2.67   26.920 32.483d 35.940 5.41   3065.00 3272.436c 3399.00 2.32
General 58.960 63.503 66.830 2.67   24.040 31.582 37.140 7.18   3059.00 3292.620 3430.00 2.78

*71 samples (09/21/2021); 78 samples (10/20/2021, 11/20/2021, 12/20/2021, 01/30/2022, 09/06/2022, 10/14/2022, 11/24/2022, 12/30/2022 and 02/06/2023); 383 samples (2021/2022); 390 samples (2022/2023); 773 samples (General).

Traits: ash (Ash, g/100 g), ether extract (EE, g/100 g), crude fiber (CF, g/100 g), starch (ST, g/100 g), amylose (AML, g/100 g), apparent metabolizable energy corrected for nitrogen balance (AMEn, kcal/kg).

Means not followed by the same letter in the column differ at the 5% significance level using the Scott-Knott test.

Among the maize genetic backgrounds (single, triple, and double cross hybrids, and open pollinated varieties), significant differences in grain composition were observed (Table 2). OPVs presented the highest Ash, EE, and CF contents, regardless of the sowing date. Meanwhile, single, triple and double cross hybrids showed the highest levels of AML, ST, and AMEn. In all genetic basis, an increase followed by a decrease in Ash, EE and CF contents was observed with the delay in the sowing date in each harvest (Table 2). For example, Ash content increased from seeding on September 21, 2021 to November 20, 2021, with subsequent reduction until January 30, 2022. Ash and CF contents were higher in the 2022/2023 harvest compared with the 2021/2022 harvest, while the highest ST, AML, and AMEn contents were observed in the 2021/2022 harvest. In all genetic basis, a reduction followed by an increase in ST, AML, and AMEn levels were observed with the delay in the sowing date, mainly in the 2022/2023 harvest.

Mean contents and coefficient of variation (CV, %) of energy nutritional traits in grains of single-cross (SH), triple-cross (TH), and double-cross (DH) hybrids and open-pollinated varieties (OPV) sown on ten sowing dates in 2021, 2022, and 2023
Date GB Ash EE CF   ST AML AMEn
Mean CV Mean CV Mean CV   Mean CV Mean CV Mean CV
09/21/2021 SH 1.107a 6.49 3.828a 6.58 2.236a 7.01   64.671a 0.89 34.545a 4.34 3360.804a 0.74
  TH 1.116a 7.55 3.830a 5.16 2.274a 6.58   64.834a 0.75 34.224a 3.11 3362.214a 0.67
  DH 1.140a 1.75 3.853a 4.32 2.287a 4.68   63.743b 1.95 33.347a 5.42 3335.667a 1.63
  OPV 1.141a 7.47 3.931a 4.13 2.333a 7.02   64.670a 0.62 32.086b 3.71 3348.375a 0.72
10/20/2021 SH 1.159a 6.59 3.939a 6.65 2.339a 7.68   64.271a 0.76 31.504a 6.33 3332.787a 0.55
  TH 1.149a 4.95 3.958a 3.85 2.299a 5.19   64.371a 0.75 31.489a 3.18 3331.000a 0.50
  DH 1.164a 5.32 4.020a 5.79 2.339a 7.22   64.746a 0.66 31.211a 2.47 3334.125a 0.62
  OPV 1.166a 8.84 4.009a 5.97 2.397a 5.21   64.594a 0.83 31.030a 2.93 3318.625a 0.50
11/20/2021 SH 1.128a 6.32 3.987b 6.89 2.301a 6.72   64.594a 0.81 33.636a 3.54 3354.745a 0.59
  TH 1.100a 8.40 3.978b 7.59 2.340a 8.08   64.618a 0.61 33.304a 4.91 3348.733a 0.68
  DH 1.145a 4.15 4.143a 4.31 2.232a 4.42   64.595a 0.58 33.986a 2.36 3364.250a 0.43
  OPV 1.174a 5.41 4.256a 3.33 2.263a 3.84   64.778a 0.42 32.460b 1.95 3335.000b 0.45
12/20/2021 SH 1.083a 8.78 3.866b 7.63 2.271a 6.50   64.882b 1.10 33.291a 3.35 3363.340a 1.15
  TH 1.033a 5.87 3.846b 6.24 2.287a 8.27   64.865b 0.51 32.730a 4.01 3365.000a 0.85
  DH 1.085a 2.96 4.176a 5.30 2.206a 8.47   65.491a 0.63 32.941a 2.85 3384.625a 0.54
  OPV 1.081a 10.93 4.040a 6.59 2.376a 7.92   65.354a 0.86 31.297b 4.52 3354.000a 1.19
01/30/2022 SH 1.116a 7.61 3.671b 8.01 2.262a 7.30   64.240a 1.06 31.909a 4.19 3347.319a 1.77
  TH 1.094a 6.56 3.766b 4.56 2.220a 5.68   64.254a 0.57 31.737a 3.74 3365.000a 0.70
  DH 1.075a 6.91 3.975a 3.67 2.149a 5.79   64.641a 0.51 32.413a 3.22 3394.125a 0.45
  OPV 1.105a 4.64 3.941a 4.00 2.243a 7.26   64.505a 0.92 30.755b 3.57 3382.625a 0.53
09/06/2022 SH 1.100a 7.58 3.892b 5.39 2.641a 9.21   64.970a 1.15 31.794a 5.11 3367.404a 0.88
  TH 1.087a 7.48 3.896b 3.35 2.629a 8.37   65.101a 1.29 31.456a 3.27 3365.467a 0.92
  DH 1.100a 7.64 4.026b 7.17 2.561a 5.36   64.900a 1.69 30.922a 3.12 3358.875a 1.22
  OPV 1.120a 7.33 4.239a 5.80 2.663a 10.65   64.735a 1.34 29.821b 2.43 3352.250a 0.92
10/14/2022 SH 1.165b 6.12 3.868a 5.62 2.658b 5.38   61.345a 0.86 29.690a 6.36 3188.085a 1.29
  TH 1.156b 8.62 3.881a 4.29 2.673b 4.87   61.232a 1.04 29.227a 3.96 3171.933a 1.15
  DH 1.165b 8.33 4.019a 6.61 2.593b 6.12   61.431a 0.82 29.064a 4.38 3165.750a 0.88
  OPV 1.244a 4.40 4.001a 5.28 2.774a 7.42   61.020a 0.95 27.473b 7.45 3123.375b 1.17
11/24/2022 SH 1.166a 8.05 3.944a 5.75 2.610a 6.46   61.468a 1.03 29.468a 5.05 3185.872a 1.27
  TH 1.136a 9.19 3.831a 4.04 2.778a 8.13   61.453a 0.92 28.131b 5.57 3146.800b 1.10
  DH 1.165a 6.73 3.987a 3.41 2.663a 6.60   61.110a 1.45 28.156b 4.34 3143.125b 1.24
  OPV 1.225a 8.67 4.054a 7.64 2.763a 7.72   61.484a 1.14 27.877b 6.38 3140.625b 1.92
12/30/2022 SH 1.121a 8.33 3.755a 7.59 2.642a 7.17   61.346a 1.48 30.463a 5.04 3189.298a 1.32
  TH 1.075a 7.44 3.745a 5.89 2.664a 9.44   61.405a 1.35 29.781a 4.98 3173.667a 1.35
  DH 1.117a 7.86 3.921a 4.60 2.550a 5.07   61.483a 1.55 30.276a 5.23 3197.375a 1.85
  OPV 1.123a 11.27 3.946a 7.28 2.632a 10.60   61.280a 0.74 28.286b 7.69 3152.875a 1.92
02/06/2023 SH 1.089a 6.56 3.591b 7.78 2.393a 10.81   63.083a 2.62 32.688a 5.09 3272.426a 2.33
  TH 1.045a 7.19 3.618b 5.70 2.292a 6.19   63.154a 2.77 32.369a 6.47 3263.667a 2.39
  DH 1.069a 7.29 3.775a 5.41 2.204a 4.22   62.855a 3.09 32.504a 4.90 3262.625a 2.55
  OPV 1.075a 8.17 3.899a 9.61 2.376a 4.64   63.545a 2.71 31.476a 5.54 3298.750a 2.18
  SH 1.123b 7.67 3.834b 7.44 2.436a 10.27   63.485a 2.63 31.893a 6.96 3296.070a 2.63
General TH 1.099c 8.07 3.835b 5.71 2.447a 10.92   63.520a 2.66 31.426a 7.16 3288.859a 2.88
  DH 1.121b 6.94 3.999a 5.68 2.384b 9.71   63.483a 2.91 31.358a 6.78 3291.280a 3.05
  OPV 1.145a 8.87 4.032a 6.53 2.482a 10.73   63.596a 2.76 30.256b 7.28 3280.650a 3.16

GB - genetic basis.

Traits: ash (Ash, g/100 g), ether extract (EE, g/100 g), crude fiber (CF, g/100 g), starch (ST, g/100 g), amylose (AML, g/100 g), apparent metabolizable energy corrected for nitrogen balance (AMEn, kcal/kg).

Means not followed by the same letter in the column differ at the 5% significance level using the Scott-Knott test.

Grain samples were grouped by crop, genetic background, sowing months, and sowing dates, capturing 56.0% of the total variability (Figure 1). The highest volumes of PSFF and PFFH occurred in the 2021/2022 harvest, compared with the 2022/2023 harvest. We observed the possibility of grouping the samples according to the sowing months occurred in the two harvests. Sowings in September, January, and February were associated with higher levels of AML, ST, and AMEn. The January and February sowings showed lower Ash, CF, and EE contents.

Data variability (n = 773 samples) represented by two principal components stratified by crops (A), maize genetic basis (B), sowing months (C), and sowing dates for contents of ash (Ash), ether extract (EE), crude fiber (CF), starch (ST), amylose (AML), apparent metabolizable energy corrected for nitrogen balance (AMEn, kcal/kg), accumulated global solar radiation from sowing to female flowering (RSFF, MJ m−2), accumulated global solar radiation from female flowering to harvest (RFFH, MJ m−2), thermal sum from sowing to female flowering (SSFF, °C day), thermal sum from female flowering to harvest (SFFH, °C day), accumulated rainfall from sowing to female flowering (PSFF, mm), and accumulated rainfall from female flowering to harvest (PFFH, mm).

The CF showed negative correlations with AML, AMEn, ST, and EE in all genetic basis, while AML, ST, and AMEn correlated positively with each other, as well as Ash and EE levels (Figure 2). The correlation analysis revealed that PFFH was negatively correlated with CF and positively correlated with AML, ST, and AMEn. In contrast, RSFF was positively correlated with EE and CF, with these relationships being more pronounced in single- and double-cross hybrids, respectively.

Pearson linear correlation network for single-cross (A), triple-cross (B), and double-cross (C) hybrids and open-pollinated varieties (D) between contents of ash (Ash), ether extract (EE), crude fiber (CF), starch (ST), amylose (AML), apparent metabolizable energy corrected for nitrogen balance (AMEn, kcal/kg), accumulated global solar radiation from sowing to female flowering (RSFF, MJ m−2), accumulated global solar radiation from female flowering to harvest (RFFH, MJ m−2), thermal sum from sowing to female flowering (SSFF, °C day), thermal sum from female flowering to harvest (SFFH, °C day), accumulated rainfall from sowing to female flowering (PSFF, mm), and accumulated rainfall from female flowering to harvest (PFFH, mm). Red lines represent negative linear correlations between traits, while green lines indicate positive linear correlations. The line thickness reflects the correlation coefficient magnitude: thin lines correspond to weaker correlations, while thicker lines indicate stronger correlations.

3.2. Influence of meteorological variables on the content of energy nutritional traits

Meteorological variables significantly explain the nutritional energy composition of maize grains (Table 3). The PFFH was the most relevant predictor, especially for ST, CF, and AMEn, being consistently selected in the best stepwise multiple linear regression models. This trend was consistent across genetic basis, with emphasis on single- and triple-cross hybrids and open-pollinated varieties. The regression models adjusted for the prediction of CF, ST, and AMEn presented the highest coefficients of determination (0.31 ≤ R2 ≤ 0.51) and cross-validation (0.25 ≤ R2cv ≤ 0.48), with low values of the root mean square error (RMSE ranging from 0.17 to 0.22 g/100g for CF, 1.26 to 1.42 g/100 g for ST, and 65.26 to 73.61 kcal/kg for AMEn). In contrast, Ash and EE contents were less responsive to meteorological variables, with R2 below 0.24. The models adjusted with the general data set showed high predictive performance, with R2 greater than 0.40 for CF, ST, and AMEn. In single-cross hybrids, for example, a 10 MJ m−2 increase in RSFF, 10 mm in PSFF and PFFH, combined with a 10 °C reduction in SSFF, resulted in increases of 0.2953 g/100 g in ST content and 14.883 kcal/kg in AMEn.

Stepwise forward multiple linear regression for predicting the content of energy nutritional traits using meteorological variables for single, triple, and double cross hybrids and open-pollinated maize varieties sown on ten sowing dates in 2021, 2022, and 2023
Model Ash EE CF ST AML AMEn
      Single hybrids    
Intercept 1.08410 2.42824 3.03656 61.02132 34.17919 3241.48585
RSFF - 0.00074 0.00013 0.00474 −0.00435 0.21467
RFFH - - −0.00037 - 0.00588 -
SSFF 0.00008 - - −0.00942 - −0.50468
SFFH - - - - −0.00350 -
PSFF - 0.00069 −0.00093 0.00601 - 0.31742
PFFH −0.00015 0.00032 −0.00093 0.00936 0.00310 0.45153
R2 0.05 0.27 0.39 0.43 0.30 0.43
R2cv 0.04 0.26 0.38 0.42 0.29 0.42
RMSE 0.08 0.24 0.19 1.26 1.85 65.26
RMSEcv 0.08 0.25 0.20 1.27 1.87 65.88
      Triple hybrids    
Intercept 1.02160 3.31402 2.64340 62.38030 30.24880 3213.15578
RSFF 0.00007 0.00031 0.00026 0.00421 −0.00307 0.24597
RFFH - - −0.00086 - 0.00835 -
SSFF - - - −0.00948 - −0.57525
SFFH - - 0.00086 - −0.00632 -
PSFF - - - 0.00520 - 0.42786
PFFH −0.00014 - −0.00118 0.00833 0.00628 0.51140
R2 0.11 0.13 0.46 0.39 0.37 0.51
R2cv 0.08 0.11 0.42 0.35 0.33 0.48
RMSE 0.08 0.20 0.20 1.32 1.79 66.17
RMSEcv 0.09 0.21 0.20 1.36 1.84 68.12
      Double hybrids    
Intercept 0.97369 - 2.31257 60.33719 33.23160 3136.15163
RSFF - - - 0.00472 - -
RFFH - - - - - 0.54774
SSFF - - 0.00040 −0.00897 −0.00401 -
SFFH 0.00021 - - - - −0.75906
PSFF - - - 0.00711 - -
PFFH - - −0.00124 0.00975 0.00813 0.38415
R2 0.19 - 0.46 0.40 0.25 0.46
R2cv 0.14 - 0.41 0.33 0.21 0.41
RMSE 0.07 - 0.17 1.42 1.82 73.47
RMSEcv 0.07 - 0.18 1.50 1.88 76.53
      Open-pollinated variety    
Intercept 0.86756 3.61972 3.15513 59.79206 25.37945 3102.89767
RSFF 0.00016 0.00023 - - - -
RFFH - - −0.00021 0.00804 0.00242 0.58378
SSFF - - - - - -
SFFH - - - −0.00949 - −0.79100
PSFF - - −0.00103 - - -
PFFH - - −0.00097 0.00747 0.00947 0.42489
R2 0.19 0.06 0.31 0.40 0.22 0.49
R2cv 0.16 0.02 0.25 0.35 0.17 0.45
RMSE 0.09 0.25 0.22 1.35 1.93 73.61
RMSEcv 0.09 0.26 0.23 1.41 2.00 76.36
      General    
Intercept 1.01892 2.66175 3.21676 61.05231 32.76465 3151.96782
RSFF - 0.00061 0.00058 0.00460 −0.00382 0.10898
RFFH 0.00003 - −0.00156 - 0.00614 0.30744
SSFF 0.00012 - −0.00114 −0.00919 - −0.24781
SFFH - - 0.00170 - −0.00377 −0.40082
PSFF - 0.00072 - 0.00618 - 0.21056
PFFH −0.00017 0.00026 −0.00068 0.00925 0.00435 0.40613
R2 0.09 0.22 0.43 0.41 0.32 0.47
R2cv 0.08 0.22 0.42 0.40 0.31 0.46
RMSE 0.08 0.24 0.19 1.30 1.87 66.44
RMSEcv 0.08 0.24 0.19 1.31 1.88 66.95

R2: adjusted coefficient of determination; R2cv: coefficient of determination of leave-one-out cross-validation; RMSE: root mean square error; RMSEcv: root mean square error of cross-validation; -: variables not selected in the final stepwise regression model.

Content of ash (Ash), ether extract (EE), crude fiber (CF), starch (ST), and amylose (AML) and apparent metabolizable energy corrected for nitrogen balance (AMEn, kcal/kg), accumulated global solar radiation from sowing to female flowering (RSFF, MJ m−2), accumulated global solar radiation from female flowering to harvest (RFFH, MJ m−2), thermal sum from sowing to female flowering (SSFF, °C day), thermal sum from female flowering to harvest (SFFH, °C day), accumulated rainfall from sowing to female flowering (PSFF, mm) and accumulated rainfall from female flowering to harvest (PFFH, mm).

Parameter estimates for all multiple linear regression models were significant by the Student's t-test to 5%.

The random forest model performed better than the stepwise regression, especially for CF, ST, AML, and AMEn (Table 4). Among the meteorological variables, PFFH was the most important predictor, showing the highest percentage increase in mean squared error (IncMSE) across genetic backgrounds and in the general model. Overall, PFFH presented IncMSE of 47.48% for CF, 65.09% for ST, 48.02 for AML, and 62.65% for AMEn, highlighting its importance in predicting these traits. The PSFF variable also showed high importance, especially for CF (IncMSE 28.62%) and ST (IncMSE 46.34%). Across genetic backgrounds and traits, R2 ranged from 0.61 (Ash in the general model) to 0.94 (AMEn in open-pollinated varieties). In the general model, the highest R2 values were observed for CF (R2 = 0.78), ST (R2 = 0.91), AML (R2 = 0.83), and AMEn (R2 = 0.92), with a RMSE of 0.12 g/100 g, 0.50 g/100 g, 0.93 g/100 g, and 26.50 kcal/kg, respectively.

Relative importance of meteorological variables (% increase in mean square error, IncMSE) for predicting energy nutritional traits estimated by the random forest algorithm for single, triple, and double cross hybrids and open-pollinated maize varieties sown on ten sowing dates in 2021, 2022, and 2023
Variable Ash EE CF ST AML AMEn
      Single hybrids    
RSFF 8.66 29.29 19.21 25.03 35.59 24.86
RFFH 17.35 21.65 21.98 28.83 27.00 30.41
SSFF 7.45 18.27 17.47 32.43 22.57 31.00
SFFH 16.85 20.62 22.91 22.33 29.29 23.78
PSFF 17.25 12.68 26.16 38.19 31.48 38.20
PFFH 21.19 15.84 44.63 53.34 39.03 57.11
R2 0.62 0.71 0.82 0.91 0.86 0.91
RMSE 0.05 0.15 0.11 0.50 0.83 26.26
R2cv 0.04 0.24 0.43 0.71 0.57 0.71
RMSEcv 0.09 0.26 0.19 0.90 1.47 46.98
      Triple hybrids    
RSFF 12.99 13.37 10.89 12.46 15.62 12.41
RFFH 10.83 18.39 15.84 26.58 19.42 25.07
SSFF 3.16 3.96 10.38 20.51 13.97 22.61
SFFH 8.95 14.77 11.53 18.32 16.82 20.10
PSFF 8.96 8.57 14.61 25.09 22.55 24.78
PFFH 11.10 8.91 34.91 32.07 23.83 30.79
R2 0.67 0.65 0.82 0.91 0.85 0.93
RMSE 0.05 0.13 0.11 0.49 0.88 24.37
R2cv 0.03 0.04 0.43 0.66 0.49 0.74
RMSEcv 0.09 0.23 0.21 0.99 1.62 48.74
      Double hybrids    
RSFF 0.86 6.46 11.44 9.66 18.09 12.01
RFFH 5.57 6.88 11.99 17.97 15.72 19.46
SSFF −1.64 −0.26 8.78 14.56 11.37 14.94
SFFH 4.38 6.31 9.26 8.73 11.75 9.12
PSFF 2.01 8.70 13.45 19.77 24.89 26.88
PFFH 2.03 7.11 33.72 26.45 24.36 26.83
R2 0.74 0.76 0.89 0.89 0.90 0.92
RMSE 0.04 0.11 0.08 0.60 0.66 27.42
R2cv 0.04 0.06 0.53 0.49 0.55 0.66
RMSEcv 0.08 0.23 0.16 1.31 1.42 58.88
      Open-pollinated variety    
RSFF 15.12 8.41 6.82 9.98 13.95 10.73
RFFH 9.14 9.80 6.51 18.01 8.07 10.96
SSFF 2.88 3.96 2.96 16.46 12.01 22.52
SFFH 9.04 11.75 8.74 11.24 7.24 6.88
PSFF 5.25 6.96 10.28 22.48 13.76 21.19
PFFH 11.70 7.06 27.85 26.89 27.46 26.29
R2 0.71 0.69 0.85 0.92 0.85 0.94
RMSE 0.05 0.15 0.10 0.50 0.85 26.08
R2cv 0.08 0.01 0.44 0.66 0.46 0.73
RMSEcv 0.10 0.28 0.20 1.02 1.62 53.50
      General    
RSFF 18.06 36.57 18.32 29.88 37.22 30.13
RFFH 25.08 26.69 23.54 34.13 31.56 35.34
SSFF 17.34 20.03 20.40 32.07 27.40 30.51
SFFH 22.79 24.18 24.02 26.77 29.61 25.52
PSFF 19.25 20.96 28.62 46.34 36.25 49.61
PFFH 32.28 24.07 47.48 65.09 48.02 62.65
R2 0.61 0.66 0.78 0.91 0.83 0.92
RMSE 0.06 0.16 0.12 0.50 0.93 26.50
R2cv 0.07 0.20 0.42 0.73 0.55 0.75
RMSEcv 0.09 0.25 0.20 0.88 1.54 45.78

R2: coefficient of determination, RMSE: root mean square error, R2cv: cross-validation coefficient of determination, RMSEcv: cross-validation root mean square error.

Content of ash (Ash), ether extract (EE), crude fiber (CF), starch (ST), and amylose (AML) and apparent metabolizable energy corrected for nitrogen balance (AMEn, kcal/kg), accumulated global solar radiation from sowing to female flowering (RSFF, MJ m−2), accumulated global solar radiation from female flowering to harvest (RFFH, MJ m−2), thermal sum from sowing to female flowering (SSFF, °C day), thermal sum from female flowering to harvest (SFFH, °C day), accumulated rainfall from sowing to female flowering (PSFF, mm) and accumulated rainfall from female flowering to harvest (PFFH, mm).

4. Discussion

The results revealed the complexity of the interactions between environmental and genetic factors in determining the nutritional energy quality of maize grains. The variability of energy nutritional traits, influenced by years, sowing dates, and types of crossing, corroborates recent studies that highlight the need for multifactorial approaches in agronomic research (Alves and Cargnelutti Filho, 2017; Martínez et al., 2017; Zelaya Arce et al., 2025). Studies have observed variation in the nutritional energy composition of maize grains between sowing dates (Martínez et al., 2017), types of endosperm (Álvarez-Iglesias et al., 2021), soil water conditions (Safian et al., 2022; Guo et al., 2023), and biotechnology events (Vidal et al., 2024).

The observed linear relationships, such as the positive correlation between AMEn, ST, and AML, and the negative correlation between CF and AMEn, ST, and AML, were consistent with the consolidated knowledge on the composition of grains and their energy value (Alves and Cargnelutti Filho, 2017). The negative relationships involving CF may be explained by its lower digestibility and contribution to grain energy value. These interrelationships, observed across genetic backgrounds, suggest a robust physiological basis underlying the accumulation of these components in the grain.

Variation in trait levels between sowing dates may be associated with the occurrence of lower rainfall in critical phases of crop reproductive development, especially during flowering and grain filling (Safian et al., 2022; Guo et al., 2023). The PFFH was the main predictor for variation in energy nutritional contents, especially for ST, CF, and AMEn. This result corroborates studies that demonstrate the critical impact of water availability during the maize reproductive phases on biomass synthesis and the translocation of photoassimilates to the grains (Nicolas et al., 1985; Chaves et al., 2009; Guo et al., 2023). The correlation revealed a negative relationship between CF and PFFH, and a positive correlation between ST and PFFH. Water deficiency during this period can limit starch synthesis and increase the fiber proportion, directly impacting AMEn (Safian et al., 2022; Guo et al., 2023).

This indicates that lower PFFH reduces the quality of maize grains, since higher CF content reduces grain digestibility. The lower starch accumulation under lower PFFH may be related to the inhibition of endosperm cell division and the decrease in the number of endosperm cells (Nicolas et al., 1985). Lower PFFH may reduce photoassimilate supplementation by inhibiting photosynthesis and carbon metabolism, reducing starch and sugar accumulation in grains (Chaves et al., 2009). These results indicate the importance of synchrony between the maize development stage and environmental conditions to maximize the accumulation of energy nutritional traits of interest.

Previous studies have shown that water stress negatively affects the activity of enzymes involved in starch biosynthesis, resulting in reduced starch deposition in grains (Jeon et al., 2010; Yang et al., 2003). In maize, the reduction in precipitation in the period 15 to 30 days after pollination severely compromised the activity of these enzymes, being correlated with the decrease in starch content in the grains (Guo et al., 2023). Our results, by showing lower ST and AMEn contents on sowing dates with lower PFFH, reinforce this relationship, suggesting that water stress during the reproductive phase may have limited starch metabolism.

The predictive capacity of the models, with coefficients of determination (R2) up to 0.51 for CF, ST, and AMEn in stepwise regression and up to 0.94 for ST in random forest, reinforces the viability of using predictive models to estimate the nutritional quality of maize based on meteorological data. The random forest model superiority in many scenarios, with greater predictive capacity and significant reduction in errors, demonstrates the potential of machine learning algorithms in analyzing complex data.

Our results provide insight into the factors associated with the nutritional and energy quality of maize grains, with emphasis on PFFH, regardless of the maize genetic background. Among the evaluated traits, ST, AML, CF, and AMEn showed greater sensitivity to meteorological variations, while Ash and EE were more stable. This information may help reduce costs and accelerate the evaluation of genotypes in breeding programs. Furthermore, such models inc may improve the accuracy of diet formulation for poultry and swine, which traditionally relies on mean reference values, without accounting for variability between genotypes and environments.

Selecting an appropriate sowing date and genotype is a strategic approach to optimizing the nutritional quality of maize. This study, which jointly evaluated the effects of different genetic backgrounds and growing seasons on the energy composition of maize grains, provides a basis for the development of genotypes adapted to environmental variability with and with high nutritional value. The models developed are specific to the conditions evaluated and may require adjustments before being applied to other agroclimatic regions. Future research should focus on the development and validation of these predictive models under different soil and climatic conditions.

 5. Conclusions

There is variation in the contents of ash, ether extract, crude fiber, starch, amylose, and apparent metabolizable energy corrected for nitrogen balance in maize grains between genetic basis and sowing dates.

At the study site, sowings in January and February present the highest levels of starch, amylose, and apparent metabolizable energy corrected for nitrogen balance, and the lowest levels of ash, ether extract, and crude fiber. Single-, triple-, and double-cross hybrids have the highest contents of starch, amylose, and apparent metabolizable energy corrected for nitrogen balance, and open pollinated varieties have the highest contents of ash, ether extract, and crude fiber.

The contents of crude fiber, starch, amylose, and apparent metabolizable energy corrected for nitrogen balance in maize grains can be predicted from meteorological variables, in all maize genetic basis. Rainfall between female flowering and the harvest point is the main meteorological variable that determines the contents of crude fiber, starch, amylose, and apparent metabolizable energy corrected for nitrogen balance, regardless of the maize genetic basis.

 Acknowledgments

To the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq - Processes 304652/2017-2 and 304878/2022-7) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) - Financing Code 001; CAPES/PROEX/88881.009488/2024-01. To the scholarship-holding students and volunteers for their help in collecting data. To the Adisseo Brasil Nutrição Animal company, especially to the CEAN Laboratory for carrying out the laboratory analyses.

 References Álvarez-Iglesias, L.; Malvar, R. A.; Garzón, R.; Rosell, C. M. and Revilla, P. 2021. Nutritional value of whole maize kernels from diverse endosperm types and effects on rheological quality. Agronomy 11:2509. https://doi.org/10.3390/agronomy11122509 Álvarez-Iglesias L. Malvar R. A. Garzón R. Rosell C. M. Revilla P. 2021 Nutritional value of whole maize kernels from diverse endosperm types and effects on rheological quality Agronomy 11 2509 https://doi.org/10.3390/agronomy11122509 Alves, B. M. and Cargnelutti Filho, A. 2017. Linear relationships between agronomic and nutritional traits in transgenic genotypes of maize. Journal of Cereal Science 76:35-41. https://doi.org/10.1016/j.jcs.2017.05.010 Alves B. M. Cargnelutti A. Filho 2017 Linear relationships between agronomic and nutritional traits in transgenic genotypes of maize Journal of Cereal Science 76 35 41 https://doi.org/10.1016/j.jcs.2017.05.010 Arnold, C. Y. 1960. Maximum-minimum temperatures as a basis for computing heat units. Proceedings of the American Society for Horticultural Science 76:682-692. Arnold C. Y. 1960 Maximum-minimum temperatures as a basis for computing heat units Proceedings of the American Society for Horticultural Science 76 682 692 Chaves, M. M.; Flexas, J. and Pinheiro, C. 2009. Photosynthesis under drought and salt stress: regulation mechanisms from whole plant to cell. Annals of Botany 103:551-560. https://doi.org/10.1093/aob/mcn125 Chaves M. M. Flexas J. Pinheiro C. 2009 Photosynthesis under drought and salt stress: regulation mechanisms from whole plant to cell Annals of Botany 103 551 560 https://doi.org/10.1093/aob/mcn125 Gonçalves, R. A.; Santos, J. P.; Tomé, P. H. F.; Pereira, R. G. F. A.; Ascheri, J. L. R. and Abreu, C. M. P. 2003. Rendimento e composição química de cultivares de milho em moagem a seco e produção de grits. Ciência e Agrotecnologia 27:643-650. https://doi.org/10.1590/S1413-70542003000300020 Gonçalves R. A. Santos J. P. Tomé P. H. F. Pereira R. G. F. A. Ascheri J. L. R. Abreu C. M. P. 2003 Rendimento e composição química de cultivares de milho em moagem a seco e produção de grits Ciência e Agrotecnologia 27 643 650 https://doi.org/10.1590/S1413-70542003000300020 Guo, J.; Qu, L.; Wang, L.; Lu, W. and Lu, D. 2023. Effects of post-silking drought stress degree on grain yield and quality of waxy maize. Journal of the Science of Food and Agriculture 103:1530-1540. https://doi.org/10.1002/jsfa.12250 Guo J. Qu L. Wang L. Lu W. Lu D. 2023 Effects of post-silking drought stress degree on grain yield and quality of waxy maize Journal of the Science of Food and Agriculture 103 1530 1540 https://doi.org/10.1002/jsfa.12250 Jeon, J.-S.; Ryoo, N.; Hahn, T.-R.; Walia, H. and Nakamura, Y. 2010. Starch biosynthesis in cereal endosperm. Plant Physiology and Biochemistry 48:383-392. https://doi.org/10.1016/j.plaphy.2010.03.006 Jeon J.-S. Ryoo N. Hahn T.-R. Walia H. Nakamura Y. 2010 Starch biosynthesis in cereal endosperm Plant Physiology and Biochemistry 48 383 392 https://doi.org/10.1016/j.plaphy.2010.03.006 Martínez, R. D.; Cirilo, A. G.; Cerrudo, A.; Andrade, F. H.; Reinoso, L.; Valentinuz, O. R.; Balbi, C. N. and Izquierdo, N. G. 2017. Changes of starch composition by postflowering environmental conditions in kernels of maize hybrids with different endosperm hardness. European Journal of Agronomy 86:71-77. https://doi.org/10.1016/j.eja.2017.04.001 Martínez R. D. Cirilo A. G. Cerrudo A. Andrade F. H. Reinoso L. Valentinuz O. R. Balbi C. N. Izquierdo N. G. 2017 Changes of starch composition by postflowering environmental conditions in kernels of maize hybrids with different endosperm hardness European Journal of Agronomy 86 71 77 https://doi.org/10.1016/j.eja.2017.04.001 Maseko, S.; van der Laan, M.; Tesfamariam, E. H.; Delport, M. and Otterman, H. 2024. Evaluating machine learning models and identifying key factors influencing spatial maize yield predictions in data intensive farm management. European Journal of Agronomy 157:127193. https://doi.org/10.1016/j.eja.2024.127193 Maseko S. van der Laan M. Tesfamariam E. H. Delport M. Otterman H. 2024 Evaluating machine learning models and identifying key factors influencing spatial maize yield predictions in data intensive farm management European Journal of Agronomy 157 127193 https://doi.org/10.1016/j.eja.2024.127193 Mehta, B. K.; Muthusamy, V.; Baveja, A.; Chauhan, H. S.; Chhabra, R.; Bhatt, V.; Chand, G.; Zunjare, R. U.; Singh, A. K. and Hossain, F. 2020. Composition analysis of lysine, tryptophan and provitamin-A during different stages of kernel development in biofortified sweet corn. Journal of Food Composition and Analysis 94:103625. https://doi.org/10.1016/j.jfca.2020.103625 Mehta B. K. Muthusamy V. Baveja A. Chauhan H. S. Chhabra R. Bhatt V. Chand G. Zunjare R. U. Singh A. K. Hossain F. 2020 Composition analysis of lysine, tryptophan and provitamin-A during different stages of kernel development in biofortified sweet corn Journal of Food Composition and Analysis 94 103625 https://doi.org/10.1016/j.jfca.2020.103625 Nicolas, M. E.; Gleadow, R. M. and Dalling, M. J. 1985. Effect of post-anthesis drought on cell division and starch accumulation in developing wheat grains. Annals of Botany 55:433-444. https://doi.org/10.1093/oxfordjournals.aob.a086922 Nicolas M. E. Gleadow R. M. Dalling M. J. 1985 Effect of post-anthesis drought on cell division and starch accumulation in developing wheat grains Annals of Botany 55 433 444 https://doi.org/10.1093/oxfordjournals.aob.a086922 Oviedo-Rondón, E. O.; Toscan, A.; Fagundes, N. S.; Vidal, J. K.; Barbi, J. and Thiery, P. 2024. Soybean meal nutrient composition, amino acid digestibility, and energy content according to the country of origin and year of harvest evaluated via NIRS. Journal of Applied Poultry Research 33:100448. https://doi.org/10.1016/j.japr.2024.100448 Oviedo-Rondón E. O. Toscan A. Fagundes N. S. Vidal J. K. Barbi J. Thiery P. 2024 Soybean meal nutrient composition, amino acid digestibility, and energy content according to the country of origin and year of harvest evaluated via NIRS Journal of Applied Poultry Research 33 100448 https://doi.org/10.1016/j.japr.2024.100448 R Core Team. 2025. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. R Core Team 2025 R: A Language and Environment for Statistical Computing R Foundation for Statistical Computing Vienna, Austria Safian, N.; Naderi, M. R.; Torabi, M.; Soleymani, A. and Salemi, H. R. 2022. Corn ( Zea mays L.) and sorghum ( Sorghum bicolor (L.) Moench) yield and nutritional quality affected by drought stress. Biocatalysis and Agricultural Biotechnology 45:102486. https://doi.org/10.1016/j.bcab.2022.102486 Safian N. Naderi M. R. Torabi M. Soleymani A. Salemi H. R. 2022 Corn ( Zea mays L.) and sorghum ( Sorghum bicolor (L.) Moench) yield and nutritional quality affected by drought stress Biocatalysis and Agricultural Biotechnology 45 102486 https://doi.org/10.1016/j.bcab.2022.102486 Sun, W.; He, Q.; Zhou, G. and Song, Y. 2025. Response of grain quality to plant growth dynamics in summer maize as influenced by sowing dates and weather factors. Industrial Crops and Products 231:121210. https://doi.org/10.1016/j.indcrop.2025.121210 Sun W. He Q. Zhou G. Song Y. 2025 Response of grain quality to plant growth dynamics in summer maize as influenced by sowing dates and weather factors Industrial Crops and Products 231 121210 https://doi.org/10.1016/j.indcrop.2025.121210 Vidal, J. K.; Simões, C. T.; Mallmann, A. O.; Tyska, D.; Pereira, H. V. and Mallmann, C. A. 2024. A three-year study on the nutritional composition and occurrence of mycotoxins of corn varieties with different transgenic events focusing on poultry nutrition. Veterinary Sciences 11:97. https://doi.org/10.3390/vetsci11020097 Vidal J. K. Simões C. T. Mallmann A. O. Tyska D. Pereira H. V. Mallmann C. A. 2024 A three-year study on the nutritional composition and occurrence of mycotoxins of corn varieties with different transgenic events focusing on poultry nutrition Veterinary Sciences 11 97 https://doi.org/10.3390/vetsci11020097 Yang, J. C.; Zhang, J. H.; Wang, Z. Q.; Zhu, Q. S. and Liu, L. J. 2003. Involvement of abscisic acid and cytokinins in the senescence and remobilization of carbon reserves in wheat subjected to water stress during grain filling. Plant Cell & Environment 26:1621-1631. https://doi.org/10.1046/j.1365-3040.2003.01081.x Yang J. C. Zhang J. H. Wang Z. Q. Zhu Q. S. Liu L. J. 2003 Involvement of abscisic acid and cytokinins in the senescence and remobilization of carbon reserves in wheat subjected to water stress during grain filling Plant Cell & Environment 26 1621 1631 https://doi.org/10.1046/j.1365-3040.2003.01081.x Zelaya Arce, M. S.; Tagliapietra, E. L. L.; Winck, J. E. M.; Alves, A. F.; Dalla Porta, F. S.; Facco, T. B.; Streck, N. A.; Soares, M. F.; Ferrão, G. E.; Debona, D.; Costa, C. H. M.; Bega, R. M.; Fochesatto, E.; Krabbe, E. L. and Zanon, A. J. 2025. Assessing genetics, biophysical, and management factors related to soybean seed protein variation in Brazil. European Journal of Agronomy 165:127541. https://doi.org/10.1016/j.eja.2025.127541 Zelaya Arce M. S. Tagliapietra E. L. L. Winck J. E. M. Alves A. F. Dalla Porta F. S. Facco T. B. Streck N. A. Soares M. F. Ferrão G. E. Debona D. Costa C. H. M. Bega R. M. Fochesatto E. Krabbe E. L. Zanon A. J. 2025 Assessing genetics, biophysical, and management factors related to soybean seed protein variation in Brazil European Journal of Agronomy 165 127541 https://doi.org/10.1016/j.eja.2025.127541

Data will be made available on request.