Animals and diets
This study complied with the ARRIVE guidelines for animal research . The experiment was performed at a commercial farm in Poland. All experimental procedures were carried out in accordance with the approval of the Local Ethical Commission for Investigations on Animals and was in line with Polish law.
A total of thirty multiparous high-yielding dairy cows (657 ± 17 kg of BW, milk yield 40 ± 9 L/d, average annual milk yield 12,000 L/cow) were randomly selected from a Polish Holstein–Friesian herd containing 120 lactating dairy cows on a commercial farm. For the thirty days prior to the experiment, the cows were fed the same diet served as a partial mixed ration (PMR) with appropriate amounts of concentrate, followed by nutrient requirements based on the milk yield. The PMR was formulated using FeedExpert software (Rovecom, Hoogeveen, the Netherlands). On day 31 of the experiment, thirty cows were selected by lactation stage: early-stage lactation (Early, days 25–100), middle-stage lactation (Middle, days 101–250), and late-stage lactation (Late, day 250 and later) and randomly allocated into three groups of ten animals each. The sample size was selected to find a difference in milk yield, with an alpha level of 0.05 and a power of 0.8 (GraphPad StatMate 2.00). A total of ten dairy cows per group (n = 10) were necessary for this, so thirty animals were used in the study. The cows from each group were kept separately in free open-stall housing with free access to water and an automated milking system (AMS; Lely Astronaut A5, Lely Industries, Maassluis, the Netherlands). Due to the limitations of running the experiment under production conditions, the cows were supervised by two workers for 24 h, who ensured that the cows returned from the AMS to the appropriate group (early, middle, and late lactation). The cows presented themselves for milking as in the period prior to the collection period (from two to six times a day, average 3.4 ± 0.89). The collection period of the experiment lasted eight days, from days 31 to 39. During this period, an appropriate total mix ration (TMR) was offered to all cows based on milk yield. TMRs were provided twice a day. The ingredients and chemical and mineral composition of TMR differing in the lactation stage are presented in Table 1.
The forage-to-concentrate ratio was altered for each lactation stage (62:38, 72:28, and 84:16, respectively, for the early, middle, and late lactation groups). Maize and alfalfa silage were used as the main forage components. The average feed intake was monitored daily for eight days for each group by weighing the total amounts offered and leftovers. Total daily amounts of TMR were divided by the number of cows in the group to calculate individual dry matter intake. Other nutrient components and mineral content were determined for each treatment using dry matter intake (Table 2). Gas production during milking at the AMS was continuously measured using an infrared methane analyzer (Servomex 4000 Series, Servomex, Jarvis Brook, UK), as previously described by Sypniewski et al. . Additionally, milk yield and body weight were recorded, and milk samples were collected. The chemical composition of the milk was analyzed using an infrared analyzer (Milko-Scan 255 A/S N). Blood samples were collected from each cow after morning milking, during routine veterinary monitoring procedures on the farm. The milk and blood were sampled at the same day, together with methane measurements. Blood samples were taken from v. abdominalis superficialis into heparinized tubes and were centrifuged at 3000 g for 10 min. All plasma samples were stored at − 20 °C for further analysis. Feed samples were collected three times during the collection period.
The TMR was chemically analyzed using the procedures of AOAC  for DM (method no. 934.01), crude protein (CP; using a Kjel-Foss Automatic 16,210 analyzer; method no. 976.05), ash (method no. 942.05), crude fiber (CF; using FOSSTecator, Fibertec System, method 962.09), and ether extract (EE; using a Soxhlet System HT analyzer; method no. 973.18). Organic matter content was calculated by subtracting the ash concentration from DM content. The nitrogen-free extract was estimated by deducting the concentrations of crude fiber, CP, EE, and ash from the DM content.
Daily milk production and the chemical composition of the milk were quantified using an infrared analyzer (Milko-Scan 255 A/S N; Foss Electric, Hillerød, Denmark). The urea concentration of the milk was determined by infrared spectrometry using a Combi Foss 6000 analyzer (Foss Electric). The energy-corrected milk yield (ECM) for milk protein and fat content was calculated according to the following equation from van Lingen et al. :
ECM (g/kg) = milk yield (kg/d) × (0.337 + 0.116 × milk fat (%) + 0.06 × milk protein (%)).
Methane and CO2 production were measured using infrared methane analyzers (Servomex 4000 Series, Servomex, Jarvis Brook, UK). The methane emission data include measured methane production (CH4, ppm), methane yield per kg of dry matter intake (CH4/DMI, ppm/kg), and methane intensity per kg of energy-corrected milk (CH4/ECM, ppm/kg). The measurement of methane has been described in detail by Sypniewski et al. . Briefly, an infrared methane analyzer was used in the AMS. Air samples were continuously collected using a gas panel. The gas samples were distributed to the inlet port of the analyzer with a flow rate of 4 L/min. Methane concentrations were measured at two-second intervals and the data were stored on a computer using software with a database system (RS 232; AnaGaz, Wrocław, Poland). Before measuring the methane concentrations of the gas samples, the analyzers were calibrated using a standard calibration gas (Multax, Zielonki-Parcela, Poland) containing 1210 ppm of methane in nitrogen gas (99.99%). MATLAB was used to identify and quantify peaks. Peaks with a height of less than 85 ppm were discarded; 85 ppm was taken as a baseline in this barn.
The mineral content of the TMR was analyzed using flame atomic absorption spectrophotometry with a double-beam atomic absorption spectrophotometer (AA-7000 Series, Shimadzu Co., Kyoto, Japan) in six replicates. The certificate reference materials of Feed LGC7173, (LGC Standards, UK) were included in each analysis to verify the accuracy of the instrument.
Plasma concentrations of Ca, Na, K, Mg, and microelements Zn, Cu, Fe were analyzed by flame atomic absorption spectrophotometry using the atomic absorption spectrophotometer (AA-7000 Series, Shimadzu Co., Kyoto, Japan). Plasma phosphorus was determined using commercial diagnostic kits (Randox, UK) on an automatic biochemical analyzer (Alizé, Lisabio, France). The direct determination of plasma Mn content was analyzed by atomic absorption spectrophotometry with electrothermal atomization (ETAAS) using the AAS with a graphite furnace (GFA-7000, Shimadzu Co., Kyoto, Japan) via the SR (high-speed self-reversal) method of background correction and pyrolytic-coated graphite tubes. The certificate reference material of lyophilized human plasma ClinCheck Control (Recipe, Munich, Germany) was included in each analysis to verify the accuracy of the instrument.
Plasma albumin content was evaluated photometrically using an ALB500 commercial kit (Erba Lachema, Czech Republic).
No animals, experimental units, or data points were excluded from the statistical analysis. The data was statistically analyzed using GraphPad Prism statistical software (version 9.0.0, GraphPad Software, San Diego, CA, USA). For multiple comparisons, one-way analysis of variance (ANOVA) was used, followed by a Tukey’s post-hoc test. Pearson’s correlation analysis was carried out between the plasma macroelement and microelement concentrations, milk chemical compositions, and methane data, to determine the relationships between the data and the strength of the putative linear association between the variables. The differences between the mean values of the lactation groups were considered to be statistically significant at P < 0.05. The values in the tables are means and pooled standard errors of the mean (SEM).