Effects of Different Sowing Dates on Nutrient and Microbiological Quality of Maize (Zea mays L.)

Abstract

The field experiment was conducted in 2016–2018 at the Department of Agronomy of the Poznań University of Life Sciences on the fields of the Research and Education Centre in Gorzyń, Złotniki branch. It was a single-factor experiment with six sowing dates of an ultra-early maize variety: A1—12 April, A2—26 April, A3—10 May, A4—24 May, A5—7 June, and A6—21 June. Seeds of the maize variety ‘Pyroxenia’ were used in the experiment. This variety is characterized by extremely early maturity (FAO 130), rapid initial development and elongation growth. Delaying the maize sowing date from A1 to A2 resulted in a 16.5% reduction in starch content in the silage dry matter, and a 14.6% increase in the ADF (Acid Detergent Fiber) fiber fraction. The difference in milk production per hectare between maize sown on date A1 and date A6 was 14,189.51 kg/ha, representing 97.1%. Delaying the maize sowing date led to an increase in the abundance of Clostridium spp. in silages, which are responsible for increased losses of dry matter, including starch. No butyric acid was detected in the silages as a final product of butyric fermentation. The low abundance of bacteria from the family Enterobacteriaceae in the silages indicated that they were well prepared. Silages prepared from maize sown at later dates were characterized by a higher abundance of undesirable mold fungi, which are responsible for dry matter losses, including starch. The coefficient of determination showed that 38.54% of the variation in silage starch content was explained by variation in mold abundance in the silage. According to the Flieg–Zimmer scale, all silages received a very good rating, regardless of maize sowing date.

1. Introduction

Maize (Zea mays L.) is frequently used as a model crop in new cultivation technologies [1]. The selection of an appropriate maize variety is considered one of the main factors influencing crop yield in modern agriculture [2,3]. Breeding progress based on deliberate genetic modifications aims to improve selected agronomic and functional traits of crop varieties. It is most often directed towards increasing yield and encompasses a range of additional characteristics that determine the economic value of varieties (EVV). These include improvements in yield quality and resistance or tolerance to various biotic factors (diseases, pests) and abiotic factors (low and high temperatures, soil quality or acidity, and insufficient or excessive rainfall). The aforementioned factors largely limit yields and also affect other specific traits that determine the agronomic and functional characteristics of the varieties [4]. New varieties should be capable of rapid regeneration once the stress has subsided. This trait is particularly important in the context of a changing climate and the increasing frequency of extreme weather events [5].

When selecting a maize variety for silage production, the cultivation region and, consequently, the maturity class of the variety should be considered. Early-maturing varieties produce lower yields than late-maturing counterparts. Silage produced from varieties with a lower FAO value has a higher energy value than silage from varieties with a higher FAO number. This is due to the higher proportion of the ear in the green fodder of early-maturing varieties. An important group also includes hybrids that remain green for a long time (“stay-green” trait). Their stems and leaves retain greenness during grain maturation, allowing for longer accumulation of nutrients and a higher concentration of dry matter in the ears [6,7]. The stems and leaves of “stay-green” hybrids show higher resistance to pathogens causing stem rot and Fusarium ear rot [8,9]. Maize is a short-day plant. Providing an adequate amount of light affects the generative development of plants [10]. As sowing is delayed, and the day length consequently increases, the plant produces more vegetative mass. At the same time, plant development and maturation are delayed. This leads to a reduction in ear yield and dry matter content of the crop.

High-quality silage has a specific pH (3.8–4.2 for maize silage; 4.2–4.6 for grass silage), high nutritional value, palatability (pleasant smell, well-preserved structure, olive color), aerobic stability (resistance to secondary fermentation), and proper microbiological quality [11]. The proper ensiling process is primarily dependent on lactic acid bacteria (Lactobacillus, Pediococcus, Lactococcus, Enterococcus, Streptococcus and Leuconostoc). Lactic acid is the main product of sugar fermentation by these bacteria, while the by-products include acetic acid, ethanol, and carbon dioxide [12]. Yeast-like fungi and molds have the greatest impact on the quality of mature silage. The growth of yeast under aerobic conditions in the presence of lactic acid increases the silage pH and promotes the proliferation of other aerobic microorganisms (spoilage), especially when the pH rises above 4.5. Mold in silage only appears in significant quantities when it has been significantly spoiled by yeasts and other aerobic bacteria. They can pose a serious threat by producing mycotoxins and causing silage breakdown under aerobic conditions [13]. Acetic acid bacteria are aerobic microorganisms that grow at low pH, oxidizing ethanol to acetic acid and, once acetic acid is depleted, converting it to carbon dioxide and water. This raises the silage pH and allows the growth of other aerobic microorganisms, thereby reducing silage quality. The families Bacillaceae, Paenibacillaceae and Enterobacteriaceae group bacteria that grow under both aerobic and anaerobic conditions. The first two families break down sugars and organic acids in silage and continue the spoilage process once the silage is exposed to oxygen. Enteric bacilli (Enterobacteriaceae), as facultative anaerobes, become active in silage under anaerobic conditions. In the initial phase of the ensiling process, enteric bacilli compete, among others, with lactic acid bacteria for nutrients. Most species, however, die or their reproduction is inhibited when the pH decreases below 4.5–5.0. It should therefore be noted that the course of fermentation processes during ensiling largely depends on the consortium of microorganisms and the metabolites they produce. It is necessary to continuously monitor the microbial communities present on the plants as well as those active during ensiling in order to produce high-quality silage. Understanding the consequences of the successive fermentation phases and correlations with microorganisms and their metabolites should be a focus of contemporary research [14].

The aim of the study was to determine the effect of different sowing dates of an ultra-early maize variety on: (i) silage chemical composition and nutritional value, (ii) silage microbiological status, and (iii) milk production. The study also presented functional relationships between milk production and selected climatic parameters, i.e., total precipitation, average daily air temperature, and the sum of effective temperatures (SET), as well as linear relationships between the starch content in the silage and the abundance of molds. In addition, linear correlation coefficients were determined for six selected silage traits.

2. Materials and Methods

2.1. Experimental Design

The field experiment was conducted by the Department of Agronomy at the Poznań University of Life Sciences on the fields of the Gorzyń Research and Education Center, Złotniki branch, in 2016–2018. The study was conducted over three years using the same design with four field replications. It was a single-factor experiment with six sowing dates of an ultra-early maize variety: A1—12 April, A2—26 April, A3—10 May, A4—24 May, A5—7 June, and A6—21 June. The same level of mineral fertilization was applied to all experimental plots: 130 kg N/ha (urea), 50 kg P₂O₅/ha (granulated triple superphosphate), and 80 kg K₂O/ha (potassium salt). Weeds were controlled after sowing (for each maize sowing date) using Lumax 557.5 SE at a rate of 4 L/ha. Maize was sown using a Monosem precision planter. The target plant density in each year of the study was 7.95 plants/m² (79,500 seeds/ha), with a row spacing of 70 cm and a sowing depth of 4–5 cm. The gross plot size was 24.5 m² (length 8.75 m, width 2.8 m). The net plot area for observations (plant sampling) was 12.25 m². The two central rows of each experimental plot were designated for maize plant sampling. Maize green fodder was harvested each year at the grain dough stage for each of the six sowing dates. Maize plants were cut into 1.0–1.5 cm pieces using a Viking GE 105 shredder (LG). After shredding, the maize silage was placed in a silo with a volume of approximately 10,600 cm³. After transferring to the laboratory, the maize samples were pre-dried under infrared lamps (250 W) for 72 h. The samples were then ground using a Retsch SM100 laboratory mill with a 0.25 mm sieve. The following components were determined in the collected samples: dry matter using the oven-weighing method with an SUP4 dryer (PN-ISO 6496:2002), crude ash using an RT-921 muffle furnace (PN 76/R-64-795), crude protein using a 2200 Kjeltec Auto Distillation Foss Tecator apparatus (PN-ISO 5983:2000), crude fat using a Soxtec HT 1043 Extraction Unit Tecator apparatus (PN-ISO 6492:2005), crude fiber using a Fibertec System 1010 Heat Extractor Tecator apparatus (PN-EN-ISO 6865:2002), fiber fractions, including neutral detergent fiber, acid detergent fiber, and acid detergent lignin, determined by the Van Soest method using an Ankom 220 Fiber Analyzer, and starch (PN-R-64785).

2.2. Silage Quality Analysis

Ammoniacal nitrogen was determined using the Conway method, while the contents of lactic, acetic, and butyric acids were measured by the Lepper method. Silage quality was assessed using a modified Flieg-Zimmer scale, and pH was measured with a CP-315 pH meter.

2.3. Silage Microbiological Analysis

The microbiological analyses of the silage included determination of the total counts of yeasts and yeast-like fungi, lactic acid bacteria, bacteria of the family Enterobacteriaceae, bacteria of the genus Clostridium, bacteria of the genus Bacillus, and molds.

2.4. Calculation of Milk Production

Milk production was calculated using equations reported by Schwab et al. [15], based on a computer program.

2.5. Climatic Conditions

The climatic conditions during the field experiment period were described using data from the meteorological station of the Research and Education Centre in Gorzyń, Złotniki branch. Thermal conditions during the maize growing season, irrespective of sowing date, were similar across the study years and averaged 15.8 °C in 2016, 14.2 °C in 2017, and 16.6 °C in 2018. Significantly greater differences between the study years were observed in total precipitation. The highest total precipitation was recorded in 2017 (553.0 mm), whereas the lowest was observed in the final year of the study, amounting to 230.3 mm (

Table 1. Average monthly air temperature and total precipitation during the growing season.

Years	Temperature [°C]
	April	May	June	July	August	September	October	Average
2016	9.6	16.3	19.9	20.3	19	17.3	8.4	15.8
2017	7.3	13.7	17.4	18.0	18.9	13.3	10.6	14.2
2018	12.9	16.9	18.5	20.2	21.3	15.8	10.9	16.6
Years	Precipitation [mm]
2016	47.3	47.3	123.8	132.8	50.3	4.6	105	511.1
2017	40.6	56.8	68.2	168.0	82.0	45.6	91.8	553.0
2018	36.2	17.4	25.6	70.5	11.6	44.2	24.8	230.3

).

In the first year of the study (2016), the highest total precipitation was recorded for the fifth sowing date (A5), accompanied by the lowest air temperature. The sum of effective temperatures (SET) for this sowing date was 1412.6 °C, while the length of the maize growing season was 125 days. The lowest total precipitation was recorded for the fourth sowing date (A4), accompanied by the highest air temperature. The sum of effective temperatures (SET) for this sowing date was 1479.7 °C, while the length of the maize growing season was 118 days (Table S1).

In the second year of the study (2017), the highest total precipitation was recorded for the last sowing date (A6), accompanied by the lowest air temperature. The sum of effective temperatures (SET) for this sowing date was 1285.6 °C, while the length of the maize growing season was 135 days. The lowest total precipitation was observed for the first sowing date (A1), with an average daily air temperature of 14.73 °C. The sum of effective temperatures (SET) for this sowing date was 100.3 °C, while the length of the maize growing period was 114 days (Table S1).

In the final year of the study (2018), the highest total precipitation was recorded for the fifth sowing date (A5), with an air temperature of 18.93 °C. The sum of effective temperatures for this sowing date was 1501.7 °C, while the length of the growing period was 116 days. The lowest total precipitation was recorded for the fourth maize sowing date (A4), with an average daily air temperature of 19.87 °C. The sum of effective temperatures (SET) for this sowing date was 1523 °C, while the length of the growing period was 110 days (Table S1).

The sums of effective temperatures (°C) were calculated based on daily temperature data from the sowing date until the maize kernels reached the black layer stage at the base. The calculations were performed using the formula reported by Weber et al. [16]:

$$\mathrm{SET}=\sum \left(\left({\mathrm{T}}_{\max }+{\mathrm{T}}_{\min }\right)/2-{\mathrm{T}}_0\right),$$

(1)

where

SET—Sum of Effective Temperatures (°C),

${\mathrm{T}}_{\max }$—maximum temperature (°C),

${\mathrm{T}}_{\min }$—minimum temperature (°C),

${\mathrm{T}}_0$—base temperature (6 °C).

2.6. Soil Conditions

The Research and Education Center in Gorzyń is located within the Poznań Upland, at an elevation of 105–110 m above sea level, in the area affected by the Baltic glaciation—Poznań Stadial. The experimental field where the study was conducted was situated on a ground-moraine plain with a light clay texture. The experiment was conducted on soil classified according to PTG [17] as follows: section—autogenic soils; order—brown-earth soils; type—grey-brown podzolic soils; subtype—typical grey-brown podzolic soils; genus—glacial clay; species—strong silty sand shallowly overlying light clay. According to the international FAO soil classification, this soil is classified as Albic Luvisols, whereas under the American system it belongs to the order Alfisols. In terms of texture, according to the international classification, it is defined as loamy sand overlying loam. It was assigned to the 4th agricultural suitability complex (very good rye soil) and to soil quality class III b. The content of basic macro- and micronutrients in the soil is presented in

Table 2. Soil nutrient content and pH before the experiment during the maize growing seasons.

Specification	Years
	2016	2017	2018
P [mg P kg−1 soil DM]	10.4	7.3	4.9
K [mg K kg−1 soil DM]	9.7	10.8	11.6
Mg [mg Mg kg−1 soil DM]	4.4	5.3	5.3
pH [in 1 mol dm−3 KCl]	4.6	5.6	5.1

and

Table 3. Soil micronutrient content and pH before the experiment during the maize growing seasons.

Specification	Years
	2016	2017	2018
Cu [mg Cu kg−1 soil DM]	2.3	2.6	1.7
Zn [mg Zn kg−1 soil DM]	18.3	9.7	9.7
Mn [mg Mn kg−1 soil DM]	260.0	90.0	170.0
Fe [mg Fe kg−1 soil DM]	950.0	510.0	703.0

.

2.7. Agronomic Conditions

Winter wheat was the preceding crop for maize in each year of the study. The basic agronomic practices were carried out according to the established schedule, on the dates presented in Table S2. Nitrogen application and soil preparation for maize sowing in each planting date were performed two days before the planned sowing date.

2.8. Statistical Methods

All calculations were performed using the STATISTICA software package (version 13.3; TIBCO Software Inc., Palo Alto, CA, USA) and Microsoft Excel. Statistical analyses were conducted at the assumed significance level of α = 0.05.

2.8.1. Split-Plot Design

In each year, the experiment followed a single-factor design, with sowing date as the only experimental factor. For the combined analysis of observations from all three years (2016–2018), a hierarchical two-level model was applied, with year treated as a higher-level factor. This approach is statistically analogous to a split-plot design, with years as first-order units and sowing dates of the maize variety ‘Pyroxenia’ as second-order units.

In the split-plot design, the mixed model of observations has the following form:

$$y_{ijk}=\mu +{\gamma }_i+{\alpha }_j+{\eta }_{ij}+{\beta }_k+{\left(\alpha \beta \right)}_{jk}+{\epsilon }_{ijk},$$

(2)

where

$y_{ijk}$—observation obtained in the i-th replication (block), for j-th level of the factor ‘years’ and the k-th level of factor A,

$\mu$—overall mean,

${\gamma }_i$—random effect of the i-th replication (block) (i = 1, 2, 3, 4),

${\alpha }_j$—fixed effect of the j-th level of the factor ‘years’ (j = 1, 2, 3),

${\beta }_k$—fixed effect of the k-th level of factor A (k = 1, 2, 3, 4, 5, 6),

${\left(\alpha \beta \right)}_{jk}$—fixed effect of the interaction between the j-th level of the factor ‘years’ and the k-th level of factor A,

${\eta }_{ij}$—random error effect of first-order units,

${\epsilon }_{ijk}$—random error effect of second-order units,

Errors are assumed to have independent normal distributions:

$${\eta }_{ij}~N\left(0, {\sigma }_{\eta }\right), {\epsilon }_{ijk}~N\left(0, {\sigma }_{\epsilon }\right).$$

(3)

When the main effects (years and sowing dates) or their interaction (year × sowing date) were found to be statistically significant in the analysis of variance (ANOVA), mean comparisons were performed using Tukey’s post hoc test. The purpose of this analysis was to identify which means differed significantly.

2.8.2. Regression

A series of regression models [18] was applied in the study, including linear regression, second-degree polynomial regression, and multiple linear regression to determine the relationships between milk production and three selected meteorological factors: mean total precipitation during the study period, mean air temperature, and mean sum of effective temperatures (SET). Additionally, the linear relationship between starch content in the silage and the number of molds in the samples was assessed. The significance of individual regression coefficients in all models was verified using analysis of variance. Model fit was assessed using the coefficient of determination (R²), which indicates the proportion of the dependent variable explained by the independent variables. Moreover, Pearson correlation coefficients were calculated to evaluate the direction of the linear statistical relationships between the tested traits.

3. Results

3.1. Silage Chemical Composition

The highest average dry matter content in the silage and nitrogen-free extractives (NFE) was recorded in 2018, while the lowest was observed in 2016 and 2017. Regarding crude ash content, the highest level was measured in 2017, while the significantly lowest was observed in 2018. Considering crude fiber content, it was shown that the highest levels in silage were recorded in 2016 and 2017, with the greatest value obtained in the final year of the study (2018).

Regardless of the year, the highest dry matter content in silage was recorded for sowing date A3, which did not differ significantly from dates A1, A2, and A4. The lowest dry matter content in silage was observed for the latest sowing date, i.e., A6. In terms of crude fat content, the highest level was recorded on date A1, while the lowest levels were recorded in sowing dates A3–A6. For starch content, the highest average levels were recorded for sowing dates A1 and A3, while the significantly lowest was observed for sowing date A6. The average dry matter content in silage was significantly influenced by the variable weather conditions during the study years and by the sowing date. In 2016, the highest dry matter content in silage was recorded for sowing dates A2, A3, A4, and A5, while the significantly lowest was observed for sowing date A6. In 2017, the highest dry matter content was obtained for sowing date A2, while the significantly lowest was observed for sowing date A6. In 2018, the highest average dry matter content in silage was recorded for sowing dates A1–A3, while the significantly lowest was observed for the latest sowing date, A6 (

Table 4. Average values for silage chemical composition by year (Y) and factor A.

Experimental Factor	Factor Level	Dry Weight [% DM]	Crude Ash [% DM]	Crude Protein [% DM]	Crude Fat [% DM]	Crude Fiber [% DM]	NFE [% DM]	Starch [% DM]
Year	2016	33.09 b	4.69 ab	8.90 a	3.49 a	20.28 a	62.65 b	36.96 a
	2017	32.76 b	4.81 a	8.83 a	3.47 a	19.63 a	63.24 b	38.06 a
	2018	39.73 a	3.83 b	8.87 a	3.55 a	17.01 b	66.69 a	37.53 a
Factor A	A1	35.81 ab	4.67 a	8.80 a	3.83 a	18.27 a	64.42 a	39.73 a
	A2	37.69 ab	4.49 a	8.77 a	3.57 ab	18.24 a	64.93 a	38.06 ab
	A3	38.48 a	4.45 a	8.85 a	3.42 b	18.43 a	64.86 a	38.88 a
	A4	35.16 ab	4.25 a	8.75 a	3.36 b	19.38 a	64.18 a	37.04 ab
	A5	34.81 b	4.26 a	8.93 a	3.43 b	19.41 a	63.97 a	37.29 ab
	A6	29.22 c	4.54 a	9.11 a	3.39 b	20.10 a	62.81 a	34.09 b

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

and

Table 5. Average values for silage chemical composition for the Y × A interaction.

Year	Factor A	Dry Weight [% DM]	Crude Ash [% DM]	Crude Protein [% DM]	Crude Fat [% DM]	Crude Fiber [% DM]	NFE [% DM]	Starch [% DM]
2016	A1	32.14 fg	5.06 a	8.60 a	3.71 a	20.12 a	62.53 a	37.70 a
	A2	35.26 def	4.66 a	8.99 a	3.91 a	18.33 a	64.12 a	38.34 a
	A3	38.18 abcde	4.53 a	8.78 a	3.40 a	20.20 a	63.11 a	39.24 a
	A4	34.08 def	4.41 a	8.81 a	3.35 a	20.13 a	63.32 a	37.33 a
	A5	33.93 def	4.46 a	9.23 a	3.17 a	20.56 a	62.59 a	37.19 a
	A6	24.97 h	5.06 a	9.00 a	3.40 a	22.32 a	60.23 a	31.95 a
2017	A1	32.88 defg	5.31 a	9.05 a	3.78 a	19.14 a	62.74 a	41.93 a
	A2	36.26 cdef	4.86 a	8.70 a	3.53 a	18.99 a	63.93 a	38.37 a
	A3	33.89 def	4.91 a	8.93 a	3.46 a	18.74 a	63.97 a	38.59 a
	A4	32.68 efg	4.58 a	8.56 a	3.26 a	20.70 a	62.91 a	36.18 a
	A5	33.23 defg	4.51 a	8.71 a	3.54 a	20.44 a	62.81 a	37.69 a
	A6	27.63 gh	4.68 a	9.05 a	3.24 a	19.78 a	63.11 a	35.57 a
2018	A1	42.40 ab	3.66 a	8.77 a	4.02 a	15.55 a	68.01 a	39.55 a
	A2	41.54 abc	3.96 a	8.63 a	3.28 a	17.41 a	66.74 a	37.46 a
	A3	43.39 a	3.91 a	8.83 a	3.42 a	16.35 a	67.50 a	38.80 a
	A4	38.72 abcd	3.78 a	8.88 a	3.47 a	17.32 a	66.31 a	37.60 a
	A5	37.28 bcdef	3.83 a	8.86 a	3.58 a	17.22 a	66.52 a	37.00 a
	A6	35.06 def	3.87 a	9.29 a	3.53 a	18.21 a	65.10 a	34.75 a

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

).

3.2. Fiber Fraction Content in Silage

The highest average ADF fiber fraction content in silage was recorded in 2016 and 2017. In contrast, the lowest average content of this fraction in silage was recorded in 2018.

Irrespective of the study year, the lowest average ADF fiber fraction content was recorded for maize sowing dates A1 and A2, while the highest was observed for the latest sowing date, A6 (

Table 6. Average values of silage fiber fraction content for years and factor A.

Experimental Factor	Factor Level	NDF [% DM]	ADF [% DM]	ADL [% DM]
Year	2016	36.75 a	22.09 a	2.44 a
	2017	36.85 a	21.71 a	2.45 a
	2018	34.85 a	17.37 b	2.31 a
Factor A	A1	34.56 a	19.23 b	2.30 a
	A2	34.72 a	19.15 b	2.32 a
	A3	36.15 a	20.35 ab	2.39 a
	A4	36.43 a	20.62 ab	2.40 a
	A5	36.42 a	20.97 ab	2.43 a
	A6	38.63 a	22.04 a	2.56 a

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

and

Table 7. Average values of silage fiber fraction content for the Y × A interaction.

Year	Factor A	NDF [% DM]	ADF [% DM]	ADL [% DM]
2016	A1	37.10 a	21.44 a	2.47 a
	A2	33.93 a	19.89 a	2.26 a
	A3	36.61 a	22.04 a	2.42 a
	A4	36.42 a	22.15 a	2.40 a
	A5	36.10 a	22.02 a	2.41 a
	A6	40.38 a	25.04 a	2.67 a
2017	A1	34.28 a	20.37 a	2.28 a
	A2	35.98 a	20.34 a	2.40 a
	A3	38.27 a	22.65 a	2.53 a
	A4	37.80 a	22.48 a	2.49 a
	A5	37.52 a	22.80 a	2.50 a
	A6	37.26 a	21.62 a	2.47 a
2018	A1	32.31 a	15.90 a	2.15 a
	A2	34.25 a	17.22 a	2.29 a
	A3	33.58 a	16.35 a	2.22 a
	A4	35.06 a	17.22 a	2.31 a
	A5	35.63 a	18.09 a	2.38 a
	A6	38.27 a	19.47 a	2.54 a

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

).

3.3. Silage Quality

In the present study, no significant effect of the year or maize sowing date on silage quality was observed. It should be noted that, for the years, sowing dates, and their interaction, silage quality was very good. The highest average reduction of N-NH₃ to total nitrogen was recorded in 2018 compared to 2016 and 2017. Considering maize sowing dates, the highest average value of the studied trait was recorded for sowing date A2, while the lowest was observed for the latest sowing date, A6 (

Table 8. Average values for silage chemical composition by year and factor A.

Experimental Factor	Factor Level	Quality According to the Flieg–Zimmer Scale	Silage Quality Classification	NH3 (g)	N-NH3/Ntotal (g)
Year	2016	91 a	Very good	0.033 a	2.47 b
	2017	90 a	Very good	0.032 a	2.47 b
	2018	93 a	Very good	0.034 a	5.77 a
Factor A	A1	92 a	Very good	0.033 a	3.38 ab
	A2	93 a	Very good	0.037 a	4.08 a
	A3	91 a	Very good	0.034 a	3.54 ab
	A4	92 a	Very good	0.031 a	3.44 ab
	A5	96 a	Very good	0.033 a	3.68 ab
	A6	95 a	Very good	0.030 a	3.27 b

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

and

Table 9. Average values for silage chemical composition for the Y × A interaction.

Year	Factor A	Quality According to the Flieg–Zimmer Scale	Silage Quality Classification	NH3 (g)	N-NH3/Ntotal (g)
2016	A1	92 a	Very good	0.031 a	2.35 a
	A2	92 a	Very good	0.038 a	2.80 a
	A3	94 a	Very good	0.037 a	2.85 a
	A4	94 a	Very good	0.030 a	2.30 a
	A5	99 a	Very good	0.032 a	2.35 a
	A6	94 a	Very good	0.029 a	2.15 a
2017	A1	90 a	Very good	0.033 a	2.45 a
	A2	93 a	Very good	0.038 a	2.90 a
	A3	87 a	Very good	0.030 a	2.30 a
	A4	87 a	Very good	0.031 a	2.45 a
	A5	94 a	Very good	0.033 a	2.55 a
	A6	89 a	Very good	0.029 a	2.15 a
2018	A1	92 a	Very good	0.035 a	5.35 a
	A2	95 a	Very good	0.035 a	6.55 a
	A3	90 a	Very good	0.035 a	5.48 a
	A4	93 a	Very good	0.033 a	5.58 a
	A5	95 a	Very good	0.033 a	6.15 a
	A6	92 a	Very good	0.032 a	5.50 a

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

).

3.4. Silage pH and Acid Content

Regardless of the study year, the lowest silage pH was recorded for sowing dates A1, A3, A4, and A5, while the highest value was observed for sowing date A6. The average silage pH was significantly influenced by variable weather conditions during the study years and by maize sowing date. In 2016, the highest average silage pH was recorded for sowing date A6, while the lowest was observed for sowing date A2. In 2017, the highest average silage pH was recorded for sowing date A6, while the lowest averages were observed for sowing dates A1, A4, and A5. In the final year of the study, varying maize sowing dates did not affect this silage parameter (

Table 10. Average silage pH and acid content by year and factor A.

Experimental Factor	Factor Level	pH	Lactic Acid [%]	Acetic Acid [%]	Butyric Acid [%]
Year	2016	3.96 a	7.10 a	2.12 a	0
	2017	3.93 a	6.68 a	2.52 a	0
	2018	3.98 a	6.69 a	2.07 a	0
Factor A	A1	3.89 b	7.64 a	2.68 a	0
	A2	3.97 ab	6.37 a	1.98 a	0
	A3	3.93 b	5.98 a	2.31 a	0
	A4	3.88 b	7.03 a	2.24 a	0
	A5	3.90 b	7.48 a	1.79 a	0
	A6	4.21 a	6.43 a	2.41 a	0

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

and

Table 11. Average silage pH and acid content for the Y × A interaction.

Year	Factor A	pH	Lactic Acid [%]	Acetic Acid [%]	Butyric Acid [%]
2016	A1	3.88 bc	7.68 a	2.72 a	0
	A2	3.83 c	6.47 a	2.29 a	0
	A3	3.97 abc	4.71 a	1.50 a	0
	A4	3.85 bc	7.75 a	2.23 a	0
	A5	3.89 bc	9.45 a	1.40 a	0
	A6	4.38 a	6.54 a	2.58 a	0
2017	A1	3.83 c	8.53 a	3.13 a	0
	A2	4.02 abc	5.73 a	1.95 a	0
	A3	3.92 abc	6.32 a	3.01 a	0
	A4	3.78 c	6.89 a	2.53 a	0
	A5	3.74 c	6.82 a	2.27 a	0
	A6	4.32 ab	5.82 a	2.23 a	0
2018	A1	3.95 abc	6.70 a	2.19 a	0
	A2	4.05 abc	6.93 a	1.70 a	0
	A3	3.91 abc	6.92 a	2.43 a	0
	A4	4.01 abc	6.44 a	1.96 a	0
	A5	4.06 abc	6.19 a	1.72 a	0
	A6	3.93 abc	6.94 a	2.43 a	0

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

).

3.5. Milk Production

Regardless of the study year, the highest average milk yield was recorded for sowing dates A1–A4, while the lowest was observed for the June sowings, i.e., A5 and A6. Regarding milk yield per 1 kg of applied nitrogen, the highest values were also recorded for sowing dates A1–A4, while the lowest were observed for A5 and A6. In the present study, milk yield was significantly influenced by variable weather conditions during the study years and by sowing date. In 2016, the highest milk yield per unit area and per 1 kg of applied mineral nitrogen was recorded for sowing dates A2 and A3, while the lowest was observed for A6. In 2017, the highest average milk yield was recorded for sowing dates A1 and A4 compared to A6. In 2018, milk yield per unit area and per 1 kg of applied mineral nitrogen was not affected by sowing date (

Table 12. Average milk yield by year and factor A.

Experimental Factor	Factor Level	Milk Production [kg/ha]	Milk Yield per 1 kg of Applied Nitrogen [kg Milk/kg Nitrogen]
Year	2016	27,729.11 a	213.30 a
	2017	26,330.41 a	202.54 a
	2018	23,764.02 a	182.80 a
Factor A	A1	28,801.71 a	221.55 a
	A2	32,609.16 a	250.84 a
	A3	29,761.37 a	228.93 a
	A4	31,669.52 a	243.61 a
	A5	18,193.12 b	139.95 b
	A6	14,612.20 b	112.40 b

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

and

Table 13. Average milk production for the Y × A interaction.

Year	Factor A	Milk Production [kg/ha]	Milk Production per 1 kg of Nitrogen [kg Milk/kg Nitrogen]
2016	A1	32,677.37 abc	251.36 abc
	A2	37,164.88 a	285.88 a
	A3	31,581.80 abcd	242.94 abcd
	A4	38,423.95 a	295.57 a
	A5	14,883.32 bcd	114.49 bcd
	A6	11,643.36 cd	89.56 cd
2017	A1	34,008.05 ab	261.60 ab
	A2	28,438.97 abcd	218.76 abcd
	A3	29,820.60 abcd	229.39 abcd
	A4	38,180.84 a	293.70 a
	A5	16,979.97 abcd	130.62 abcd
	A6	10,554.01 d	81.18 d
2018	A1	19,719.72 abcd	151.69 abcd
	A2	32,223.62 abc	247.87 abc
	A3	27,881.70 abcd	214.47 abcd
	A4	18,403.78 abcd	141.57 abcd
	A5	22,716.06 abcd	174.74 abcd
	A6	21,639.22 abcd	166.46 abcd

Statistical significance was set at p < 0.05. Values sharing the same letter are not significantly different.

).

3.6. Functional Relationships Between Milk Production and Meteorological Data

Table 14. Functional relationships between milk production and total precipitation, air temperature, and SET.

Year	Correlation Coefficient r	Regression Equation	Determination Coefficient R2	p-Value
2016	−0.7821	$y=192125.02-534.79x_1$	61.18%	0.0026
	0.7534	$y=-68928.22+5882.38x_2$	56.76%	0.0047
2017	–	$y=-64019.35+12965.55x_2-95.36x_3$	69.74%	$p_2=0.0017$ $p_3=0.0024$
	−0.5923	$y=135174-287.14x_1$	35.08%	0.0424
2018	–	$y=-751303.26+1141.15x_3-0.42x_3^2$	43.86%	0.0264
Independently of the year	–	$y=-36426.32+48.68x_1+6073.08x_2-41.01x_3$	36.29%	$p_1=0.0114$ $p_2=0.0002$ $p_3=0.0047$

$y$—milk production, $x_1$—precipitation, $x_2$—temperature, $x_3$—SET.

presents the functional relationships between milk production (y) and the average total precipitation ($x_1$), average air temperature ($x_2$) and average sum of effective temperatures (SET) ($x_3$) over the study period. Analysis of the 2016 data showed a significant positive linear correlation between milk production and air temperature, indicating that higher temperatures promoted increased milk production. The coefficient of determination of 56.76% shows that air temperature accounted for nearly 57% of the variability in milk production, according to the regression model $y=-68928.22+5882.38x_2$. A significant negative linear correlation was also observed between milk production and total precipitation, suggesting that higher precipitation was associated with a decrease in milk production. The coefficient of determination of 61.18% indicates that total precipitation accounted for over 61% of the variability in milk production, according to the linear regression equation $y=192125.02-534.79x_1$. A similar relationship was observed in 2017, with a significant negative linear correlation between milk production and total precipitation. The coefficient of determination of 35.08% indicates that precipitation accounted for 35.08% of the variation in milk production, according to the corresponding equation presented in Table 14. Additionally, the 2017 analysis confirmed that air temperature and SET had a significant effect on milk production. The coefficient of determination of 69.74% indicates that the regression model accounted for nearly 70% of the variation in milk production. In 2018, a significant second-degree nonlinear relationship was observed between milk production and SET at α = 0.05. Analysis of data from 2016–2018 showed that air temperature, total precipitation, and SET had a significant effect on milk production. The coefficient of determination R² = 36.29% indicates that the obtained multiple regression model explained over 36% of the variation in milk production during the study period.

3.7. Linear Relationships Between Selected Silage Characteristics

The study examined whether linear relationships existed between the contents of six selected maize silage parameters, analyzed separately for individual study years and combined, irrespective of the year.

Analysis of all observations collected between 2016 and 2018 showed significant negative linear correlations between crude fiber and NFE, crude fiber and starch, and crude ash and NFE. Thus, the higher the crude fiber content in maize silage, the lower the starch and NFE content. Moreover, as the crude ash content in silage increased, a decrease in NFE was observed. Research also confirmed the presence of positive linear relationships between crude fiber and crude ash, and between starch and crude fat. This indicates that an increase in crude fiber content in silage led to an increase in crude ash content. It was also shown that a higher starch content in silage caused a concurrent increase in crude fat (Table S3).

In 2016 and 2018, it was demonstrated that there was a positive linear relationship between the starch content in maize silage and the quantity of NFE. It was also noted that an increase in starch content led to a simultaneous elevation of NFE (Table S4). Research conducted in 2016 demonstrated the existence of a negative linear relationship between the crude fiber content in maize silage and the amounts of crude fat, starch, and NFE. Thus, increasing crude fiber content in the silage reduced the concentrations of crude fat, starch, and NFE. Additionally, it was shown that an increase in the crude ash content in the silage resulted in a simultaneous decrease in NFE. Research conducted in 2017 confirmed a positive linear relationship between starch content and crude protein level, as well as between starch content and crude fat. It was found that raising the starch content in maize silage simultaneously increased the amount of crude protein and crude fat. The research also indicated that higher crude fiber content corresponded to lower NFE levels. The results of studies conducted in 2018 showed that there is a negative relationship between the crude fiber content in maize silage and the content of starch and NFE. Therefore, higher crude fiber content in the silage was associated with lower starch and NFE content. Moreover, as the crude fat content in the silage increased, a decrease in the amount of crude ash was observed.

3.8. Hygienic Value of Silage

The results of microbiological analyses of silage samples are presented in Figures S1–S6. Figures S1 and S3–S5 were prepared using a logarithmic (log) scale, which allowed for a focus on the dynamics of changes between the factors under study. Figures S2–S6 were prepared using a linear scale due to the lower variation in the abundance of microorganisms.

3.8.1. Lactic Acid Bacteria

Analysis of the growth dynamics of lactic acid bacteria over time demonstrated a common trend for both 2017 and 2018. Silages prepared from maize sown earlier (A1–A3) had higher counts of these bacteria, whereas their populations were markedly smaller in silages from later sowing dates. This relationship was consistent across all sampling depths (A, B, C).

At level A in 2017, the highest abundance of lactic acid bacteria was recorded at sowing date A2, reaching log 8.52 CFU/g silage DM, while the lowest was at sowing date A6, at log 4.7. The abundance of lactic acid bacteria from sowing dates A2 to A6 decreased by 42.7%. The highest abundance of lactic acid bacteria in 2018 was recorded at sowing date A1, reaching log 8.9. The lowest values were recorded at sowing date A6, reaching log 4.4. The abundance of these bacteria decreased by 50.6% from sowing dates A1 to A6 in 2018. At level B, the highest abundance of lactic acid bacteria was recorded at sowing date A1, reaching log 8.7 in both 2017 and 2018. In 2017, the lowest value, log 4.0, was recorded at sowing date A6, while in 2018, it was recorded at sowing date A5, reaching log 3.0. The population of these bacteria decreased by 54% in 2017 and by 65.5% in 2018. At sowing date A6, bacterial abundance increased by 43% compared with date A5. At level C, the highest counts were recorded at sowing date A1 in 2017 and 2018, log 8.9 and log 8.5, respectively. In 2017, the lowest value was recorded at sowing date A6—log 4.3, while in 2018, it was at sowing date A5—log 4.8, which means that their abundance decreased by 52% and 43.5%, respectively. In 2018, the proportion of lactic acid bacteria increased by 16.4% from sowing dates A5 to A6 (Figure S1).

3.8.2. Bacteria from the Family Enterobacteriaceae

Analysis of the growth dynamics of lactic acid bacteria over time showed a common trend for both years, 2017 and 2018. Silages made from maize sown at earlier dates (A1–A3) were characterized by lower counts of these bacteria, while their abundance was significantly higher for later maize sowing dates. This relationship was consistent across all sampling depths (A, C, excluding level B for 2017). A distinctly higher count of these microorganisms was observed in 2017, while in 2018, these bacteria were either undetectable or present at levels near the detection limit at sowing dates A1–A3. At level A in 2017, the abundance of Enterobacteriaceae decreased across sowing dates A1–A3 as follows: 10 × 10², 19 × 10², 0.0 × 10² CFU/g. At the same sowing dates in 2018, these bacteria were not detected using the methods applied in the present study. From sowing date A3 onwards, an increase in Enterobacteriaceae abundance was observed at sowing dates A4–A6 in both 2017 and 2018. In 2017 and 2018, the highest bacterial abundance was recorded at sowing date A6, reaching: 44 × 10², 110 × 10² CFU/g. In 2017, level B was characterized by a high abundance of Enterobacteriaceae, amounting to 82 × 10² CFU/g. Their number decreased in subsequent sowing dates and reached a value of 0.0 × 10² CFU/g at sowing date A3. In 2018, the abundance of these bacteria at sowing dates A1–A3 was recorded at 0.0 × 10² CFU/g. In both 2017 and 2018, an increase in the number of these bacteria was observed from sowing date A3 onwards. In 2017, the highest abundance, 83 × 10² CFU/g, was recorded at sowing date A4, after which it decreased. In 2018, their counts at sowing date A6 reached its highest value of 220 × 10² CFU/g. The lowest Enterobacteriaceae abundance at level C in 2017 was recorded at sowing date A3, at 0.0 × 10² CFU/g. At the same level in 2018, the number of these bacteria was recorded at the same level for sowing dates A1–A2. In both study years, an increase in Enterobacteriaceae abundance was observed from sowing date A3 onwards, reaching maximum values at sowing date A6. These amounted to 180 × 10² and 290 × 10² CFU/g in 2017 and 2018, respectively (Figure S2).

3.8.3. Bacteria from the Genus Clostridium

The developmental dynamics of this group of microorganisms were similar in 2017 and 2018. Silages prepared from maize sown at earlier dates (A1–A3) had a lower abundance of these bacteria, whereas their counts were markedly higher in silages from later sowing dates. This relationship was consistent across all sampling depths (A, B, C). At level A, lower counts of these bacteria were recorded in 2017. The lowest values were obtained for sowing dates A1–A4 and amounted to log 2.7, log 2.0, log 2.6, and log 2.1, respectively. With delayed sowing dates, the abundance of these bacteria increased and reached a maximum at sowing date A6, at log 5.3, representing an increase of 62.3%. In 2018, the abundance of Clostridium sp. at sowing dates A1–A3 was at a similar level and represented the lowest values, at log 2.0, log 2.6, and log 2.7, respectively. At sowing dates A4–A6, an increase in the abundance of these bacteria was observed, reaching a maximum at sowing date A6 of log 7.6, which was 74% higher than the initial value. Level B showed a similar trend in bacterial development. The lowest values in both study years were recorded at sowing dates A1–A3, amounting to log 2.4, log 2.0, and log 2.8 in 2017, and log 2.7, log 2.8, and log 3.0 in 2018, respectively. The highest values were recorded in 2017 and 2018 at sowing date A6, reaching log 5.7 and log 6.1, respectively, representing increases of 65% and 56% compared with the lowest values. At level C, the situation was similar for sowing dates A1–A3. The abundance of the bacteria analyzed was determined at similar levels, amounting to log 2.3, log 2.1, and log 2.3 in 2017, and log 2.3, log 2.3, and log 2.7 in 2018, respectively. In 2017, the highest number of these bacteria was recorded at sowing date A6, reaching log 6.9, which was 67% higher than the lowest values. In 2018, the highest values were recorded at sowing date A5—log 6.7, representing an increase of 66% over the initial values. Subsequently, at sowing date A6, the bacterial population decreased by 10.4% to log 6.0 (Figure S3).

3.8.4. Total Abundance of the Genus Bacillus

The growth dynamics of the genus Bacillus remained at a similar level in 2017 and 2018. The lowest abundance of these bacteria was recorded at the early maize sowing dates, i.e., A1–A3. Later sowing dates showed higher counts of Bacillus sp. It should be noted that in 2018, the population of these bacteria was larger. At level A, the lowest values were recorded at sowing date A1 in 2017, at log 2.6, and at sowing date A2 in 2018, at log 2.3. In 2017, the highest abundance of Bacillus sp. was recorded at sowing date A5—log 5.9, representing a 56% increase over the initial value. In 2018, the highest abundance of these bacteria was recorded at sowing date A6—log 8.0, representing a 71.3% increase compared with sowing date A2. At levels B and C, a significant increase in the abundance of these bacteria was observed from sowing date A3 in both 2017 and 2018. At level B, the maximum abundance was recorded at sowing date A6, reaching log 7.2 in 2017 and log 8.5 in 2018, representing increases of 62.5% and 76.4% over the lowest values, respectively. At level C, a similar trend was observed, with the highest values recorded at sowing date A6 in 2017 and 2018, at log 7.1 and log 8.9, respectively. These values were higher than the lowest values by 66.2% and 76.4%, respectively (Figure S4).

3.8.5. Total Abundance of Molds

The results showed increases in the abundance of molds in the silages from sowing date A3 in 2017 and 2018. The largest population size of these fungi was recorded at sowing date A6. The year 2018 was characterized by higher abundance of molds compared to 2017. At level A, a high number of molds was recorded at sowing date A1, amounting to log 6.3 in 2018. However, at sowing date A2, the abundance was similar in both 2017 and 2018, i.e., log 4.5 and log 4.1, respectively. The lowest values were recorded in 2017 and 2018 at sowing date A3, reaching log 4.2 and log 4.0, respectively. Sowing date A6 was characterized by the highest counts of these fungi, reaching log 8.0 and log 7.6 in 2017 and 2018, respectively, representing increases of 47.5% and 47% compared to sowing date A3. At level B, the lowest numbers of molds in 2017 and 2018 were recorded at sowing date A1, both at log 4.4. On the other hand, the highest values were recorded in 2017 and 2018 at sowing date A6, reaching log 7.4 and log 7.0, respectively; they were higher than the lowest values by 40.5% and 37%, respectively. Level C showed a similar growth pattern of molds compared with level B. The lowest abundance was recorded at sowing date A1 in 2017 and 2018, both at log 4.4. The highest values were recorded at sowing date A6, reaching log 7.9 and log 7.7, representing increases of 44.3% and 42.9% over the lowest values, respectively (Figure S5).

3.8.6. Total Abundance of Yeasts and Yeast-like Fungi

Analysis of the results for the total abundance of yeasts and yeast-like fungi did not reveal a strong trend in their growth dynamics in 2017 and 2018. Similarly, their development at the individual sampling levels do not allow to draw clear conclusions. The lowest abundance of these fungi was 19 × 10³ CFU/g, recorded in 2018 at sowing date A1 for level B, and at sowing date A3 for level C. The highest values were 97 × 10³ CFU/g, recorded at level A at sowing date A4 and at level C at sowing date A3 (Figure S6).

4. Discussion

Harvesting plants at the optimal growth stage is the period when nutrient concentrations in the plants are high and the content of water-soluble sugars reaches its maximum, allowing lactic acid bacteria to carry out proper fermentation. Maize suitability for ensiling is assessed based on the grain. Grain from the middle section of the ear should have a milk–dough or vitreous maturity, with the milk line running halfway to two-thirds from the kernel base [19]. Each day of delayed harvest not only results in nutrient losses and reduced digestibility but also slows down the pH reduction of the ensiled mass, promoting the growth of undesirable microorganisms. In the present study, for each maize sowing date, harvesting was carried out when a black spot appeared at the base of the kernels in the middle section of the ear.

Dry matter content is one of the main factors determining proper ensiling and high silage quality. Dry matter content of the ensiled material should range from 30 to 45%, ensuring a high concentration of sugars and optimal conditions for lactic acid production. Mowing combined with harvesting and subsequent ensiling is characteristic of maize silage. Silages prepared from forage with optimal dry matter content have higher nutritional value and are more readily consumed by animals [20,21,22]. Dry matter content of the ensiled material above 50% is undesirable, as it leads to high mass compressibility and difficulties in removing air. Excessive drying inhibits the development of lactic acid bacteria, preventing the pH from reaching the desired level. Silages with high dry matter content have higher pH, lower concentrations of acetic and propionic acids, and poor aerobic stability. Undesirable microorganisms from the genera Bacillus, Salmonella, Listeria, as well as yeasts and molds, can develop in such silages [23]. Ensiling plants with low dry matter content (below 25%) is also problematic, as it leads to the leakage of silage juices rich in nutrients, particularly simple sugars essential for the development of lactic acid bacteria [24]. Low dry matter content in the ensiled mass creates conditions favorable for the growth of the genus Clostridium. Saccharolytic clostridia carry out butyric fermentation, whereas proteolytic strains break down proteins and nitrogen compounds into ammonia and biogenic amines. Such silage has an unpleasant odor, high acetic acid content (3–4% dry matter), the presence of butyric acid, and is poorly consumed by animals [23,25]. Dry matter losses during feed fermentation mainly depend on the types of microorganisms developing in the plant mass and the availability of specific substrates necessary for proper ensiling. It should be noted that some fermentation products have higher energy content compared to their substrates, resulting in considerably greater dry matter losses than gross energy losses [26]. As reported by Wróbel et al. [27], losses caused by lactic acid bacteria depend on the bacterial group carrying out a specific type of fermentation. Under anaerobic conditions, homofermentative lactic acid bacteria primarily synthesize lactic acid from glucose or fructose. This process occurs without dry matter loss, as no gaseous products are formed, and gross energy loss is only 0.7%. In contrast, glucose fermentation by heterofermentative lactic acid bacteria results in a 24% loss of dry matter, while gross energy loss amounts to only 1.7%. Heterofermentation is associated with relatively high dry matter losses, amounting to 24%, because 0.52 mol of CO₂ is produced from one mole of lactic acid [28]. According to Irawan et al. [29], dry matter losses in whole-plant maize silages range from 22.2 to 27.4%. Dry matter losses caused by bacteria from the family Enterobacteriaceae are primarily due to their competition with lactic acid bacteria for simple sugars, and they are also capable of protein degradation. The resulting ammonia increases silage buffering capacity, preventing a rapid decrease in pH. Glucose fermentation by Escherichia coli produces two moles of lactic acid, one mole of acetic acid, and two moles of CO₂. Fermentation carried out by another member of this family, bacteria of the genus Klebsiella, can lead to greater dry matter losses in silages. From two moles of glucose, only one mole of lactic acid is produced, and the mass loss is 1.5-fold higher than that resulting from glucose fermentation by E. coli [30]. Clostridia are capable of breaking down substances such as carbohydrates, proteins, amino acids, amides, and lactic acid, using them as energy sources. The products of their fermentation typically include butyric acid, acetic acid, NH₃, CO₂, and H₂. Butyric acid is weaker than lactic acid, and its presence in silage raises the pH. Moreover, two moles of lactic acid yield only one mole of butyric acid. Thus, losses of the preserved material occur due to the formation of gaseous fermentation products. These account for 51.1% in the fermentation of glucose and lactic acid to butyric acid, CO₂, and H₂ [27]. Considering that dry matter losses in silages resulting from the activity of clostridia are very high, efforts should be made to reduce their abundance [31].

Under anaerobic conditions, many yeast species ferment sugars such as glucose, maltose, and sucrose, producing ethanol and CO₂, while small quantities of other metabolites, including butyric, propionic, and acetic acids, are also formed during these processes [32]. Conversion of glucose and lactic acid to ethanol and CO₂ can lead to dry matter losses of up to 49% and an increase in the pH of the ensiled feed [24,25]. Borreani et al. [33] demonstrated a relationship between dry matter losses and mold abundance in the silage. When mold levels in silage exceeded 5 log₁₀ CFU/g (mold became visible on the silage), dry matter losses exceeded 20%. When mold counts exceeded 6.0 log₁₀ CFU/g silage, losses could exceed 40% of the originally ensiled dry matter. Moreover, when mold counts increased to more than 5.0 log₁₀ CFU/g silage, significant deterioration in nutritive value was observed, with starch content beginning to decline and falling below 10% dry matter when mold counts exceeded 7.0 log₁₀ CFU/g silage. As silages mature, their chemical and microbiological composition substantially changes, thus feeds ensiled from for ages with different chemical profiles may differ in the quality of their fermentation processes [34]. Major errors in silage preparation can be related to incorrect harvest timing [35,36], plant developmental stage, and diurnal cycle [37], which may reduce the rate of acidification of the plant material and increase dry matter and energy losses. The microbiome of ensiled forages reflects the microbiota of the soil from which they were harvested. This, in turn, depends on the geographic region, soil quality, fertilization, local animal populations (insects, rodents, birds), and climatic conditions. The plant phyllosphere microbiome also undergoes significant quantitative and qualitative changes depending on forage harvest time for silage [26]. It is essential to continuously study the microbial communities present on plants as well as those active during ensiling to produce high-quality silage [37]. Plants intended for ensiling host aerobic and anaerobic bacteria, as well as fungi, which directly influence the quality of the resulting silage [24]. Proper silage preparation should promote the growth of lactic acid bacteria and minimize losses of sugars and other nutrients [23].

Drouin et al. [38] observed that the bacterial communities introduced with plant material into silos can also differ significantly in abundance depending on the plant harvest time. In the analyzed silage samples, the 2017 and 2018 crops had higher dry matter content in maize straw and ears at harvest dates A1–A3, which may have led to higher abundance of lactic acid bacteria, while lower abundance was recorded at harvest dates A5–A6. According to da Silva et al. [39], most lactic acid bacteria introduced into the silo are not osmotolerant. On the other hand, it should be noted that the number of these bacteria can decline sharply with delayed harvest (within the diurnal cycle), potentially due to UV radiation exposure [40]. In the present study, the total lactic acid bacterial counts ranged from 10⁴ to 10⁸ CFU/g silage, which was consistent with the results reported by Selwet [41] for maize silage at 10⁶ CFU/g silage, and by Selwet et al. [41] at 10⁸ CFU/g silage. On the other hand, low dry matter content can promote the growth of anaerobic bacteria of the genus Clostridium, leading to butyric fermentation [42]. The genus Clostridium can be divided into two main groups causing nutrient losses in silage: saccharolytic species, which drive butyric fermentation (Cl. butyricum, Cl. parabutyricum, Cl. tyrobutyricum, and Cl. scatol) and proteolytic species, which degrade proteins (Cl. sporogenes, Cl. perfringens, and Cl. botulinum) [43]. The conversion of lactic acid to butyric acid is one of the most energy-intensive reactions, resulting in a loss of approximately 50% of silage dry matter [44]. Analysis of silage samples from 2017 and 2018 showed an increase in Clostridium abundance at harvest dates A3–A6. Ensiling maize that is too moist promotes the growth of these bacteria. However, no decrease in protein concentration or increase in N-NH₃ and butyric acid concentrations were observed at these harvest dates. Low protein degradation and low ammonia nitrogen concentration may result from a rapid pH decrease and inhibition of fermentation carried out by Clostridia [45]. In our study, the total abundance of the genus Clostridium ranged from 10² to 10⁷ CFU/g silage, which was consistent with the results reported by Selwet et al. [46] for maize silage at 10² CFU/g. Bacillus species are capable of fermenting sugars and synthesizing organic acids in silage, and can initiate spoilage when the silage is exposed to oxygen; some, such as Bacillus cereus, are also pathogenic (BdUth1) [47,48]. In 2017 and 2018, the lowest counts of bacteria from the genus Bacillus were recorded at harvest dates A1–A3. At the same time, the highest abundance of lactic acid bacteria was also determined at these dates, which may have been associated with the production of lactic and acetic acids at levels sufficient to limit the growth of these bacteria. According to Liu et al. [49], rapid silage acidification depends on lactic acid production, which is essential to prevent the early proliferation of undesirable microorganisms and nutrient losses. In our study, the highest Bacillus abundance was recorded at later harvest dates, A4–A6. According to Dong et al. [40], delayed harvest of sorghum plants led to an increase in the proportion of these bacteria in silage. At the same time, Ogunade et al. [50] reported that the number of endospores of these bacteria in maize silage ranged from 10² to 10⁶ CFU/g and was significantly lower in deeper layers of the silo. This was not confirmed by our study, where Bacillus counts ranged from 10² to 10⁸ CFU/g silage, regardless of sample depth. Excessively high dry matter content (>50%) increases the risk of fungal infection, including toxin-producing species. An important aspect of silage quality is its aerobic stability, i.e., its resistance to oxidative spoilage. The direct cause of silage aerobic instability is the growth of yeasts and molds. The activity of molds and yeasts in ensiled plant material results from excessive wilting of the forage [51]. However, da Silva et al. [39] reported higher abundance of these fungi in maize silage with low dry matter content. This, as the authors explain, may have been due to higher moisture content and water-soluble sugars. Muck [52] observed only slight differences in yeast abundance in whole-plant maize silage with varying dry matter content, reporting 10⁶ CFU/g at 69.2% DM and 10⁵ CFU/g at 34.7% DM. The results of our study indicate low and variable counts of yeasts and yeast-like fungi at individual harvest dates (10³ CFU/g). The development of these fungi may have been influenced by competition for fermentation substrates with lactic acid bacteria and Enterobacteriaceae [21]. Anaerobic conditions, low pH, and the presence of acetic and propionic acids create an unfavorable environment for yeast growth. Undissociated acid molecules more readily enter yeast cells and, through the release of H⁺ ions, acidify the cells, leading to their death [27]. Aerobic deterioration in silage begins as early as the first day of ensiling and is closely associated with the degree of material compaction and the time interval between bale or bunker formation and sealing [53]. Aerobic processes during silage storage result from increased oxygen infiltration and are associated with mold growth on the surface [54]. In the current study, the total counts of lactic acid bacteria ranged from 10⁴ to 10⁸ CFU/g silage, which was consistent with the values reported by Selwet [41] for maize silage at 10⁵ CFU/g silage but higher than those reported by Selwet et al. [46] at 10² CFU/g silage. Microbiological assessment also takes into account the abundance of potentially pathogenic Enterobacteriaceae, including Escherichia coli and Salmonella [55]. In well-prepared silages, the abundance of Enterobacteriaceae is generally very low, often at the detection limit [36], which was confirmed by the results of this study at harvest dates A1–A3, particularly in 2018. Additionally, the low pH values of the silages may have limited the abundance of these bacteria, particularly during the early ensiling stages [25]. In the present study, the total counts of Enterobacteriaceae were at a level of 10² CFU/g silage, whereas higher counts in maize silage were reported by Selwet et al. [46] at 10⁵ CFU/g silage. The study investigated the linear relationship between starch content in silage (characteristic Y) and mold abundance in silage samples (characteristic X). Analysis of all observations collected in 2017–2018 showed the presence of a negative linear relationship between these factors. Thus, higher mold counts were associated with lower starch content in the silage. The coefficient of determination (R² = 38.54%) indicated that 38.54% of the variability in silage starch content was explained by variation in mold counts, according to the linear regression model presented in Table S5. Analysis of the data collected in all study years for depth A showed that there was also a negative linear relationship between silage starch content and the mold counts. The coefficient of determination (R²) of 36.70% indicated that over 36% of the variability in silage starch content was influenced by variation in mold counts, according to the regression equation presented in Table S5. Furthermore, a negative linear relationship between the characteristics was observed for data collected in 2018 at depth A. The coefficient of determination (R² = 74.21%) indicates that over 74% of the variation is explained by the fitted linear regression function. Approximately 26% of the variation in characteristic Y (silage starch content) was not explained by the variation in characteristic X (mold abundance in silage).

Grass and legume silages, or their mixtures, contain no starch and only small amounts of sugar. In contrast, silages from whole plants (WPs) and maize contain starch at levels ranging from approximately 30% to nearly 70% dry matter. The quality of the fermentation process is indicated by the content of residual sugars that were not converted into lactic acid during fermentation. Its level should be below 10% of the silage dry matter. High sugar and starch content promotes the growth of yeasts and molds and indicates low silage resistance to heating during feed-out and feeding. The fewer yeasts and molds are present, the more the silage remains stable after opening the silo [56]. An important source of energy for ruminants is starch from maize ears, including its level, ruminal degradation, and digestibility in the small intestine. In properly prepared silages, starch content should exceed 30% per kilogram dry matter [57]. Plants harvested later, with higher dry matter content, contain more starch [58]. The threshold number of molds in silages of good hygienic quality is 4 log CFU g⁻¹ silage [13]. Silage spoilage by fungi is associated with nutrient and dry matter losses, reduced palatability, mycotoxin production, and decreased feed intake [59]. Large differences in silage quality and dry matter losses may not be reflected in significant differences in gross energy content. Therefore, one of the main objectives of ensiling is to minimize dry matter losses during the preservation process. In recent years, numerous studies have been conducted in many countries aimed at reducing these losses. A key prerequisite is understanding the interactions between microbial groups growing in the ensiled material. On this basis, certain good practices for silage management have been developed to prevent or at least minimize feed losses [27]. Molds can be present in silage at all stages of fermentation. These fungi hydrolyze carbohydrates, proteins, and organic acids, thereby reducing silage acidity [60]. Therefore, during ensiling, molds are often found in poorly compacted areas, such as the top layer and the sides of silos, where exposure to air is highest [50]. In studies on maize silage conducted by Borreani et al. [33], total mold counts were negatively correlated with starch content. Starch content in silage began to decline when mold abundance increased >5 log₁₀ CFU/g silage and decreased below 10% dry matter when mold counts exceeded >7 log₁₀ CFU/g silage.

5. Conclusions

A delayed maize sowing date led to a higher abundance of Clostridium species in silages, which are responsible for increased dry matter losses, including starch. Butyric acid was not detected in the silages as a final product of butyric fermentation. The low abundance of bacteria from the family Enterobacteriaceae in the silages indicated that they were well prepared. Silages prepared from maize sown at later dates were characterized by a higher abundance of undesirable molds, which also contribute to dry matter losses, including starch. The coefficient of determination showed that more than 38.54% of the variation in silage starch content was explained by differences in mold abundance in the silage. According to the Flieg–Zimmer scale, all silages, irrespective of the maize sowing date, received a very good rating.