Validation and selection of nutritional markers to evaluate mixing quality in total mixed ration for dairy cows: Technical note

Highlights

• •Magnesium identified as the most consistent TMR mixing quality marker.
• •NDF and CP also reliable indicators of TMR homogeneity.
• •Recommended CV values: Mg (9.5%), NDF (9.0%), CP (9.3%).
• •Proposed method improves accuracy in monitoring TMR uniformity.

Abstract

Dairy cow diets can be categorized into three types: the formulated diet, the farm-prepared diet, and the diet consumed by cows. Aligning these diets requires monitoring the quality of the total mixed ration (TMR) to ensure consistency. However, limited information exists on nutritional markers to assess TMR homogeneity. We hypothesized that nutritional markers could monitor TMR mixing quality in ruminant diets. Therefore, this study aimed to validate a methodology, determine an appropriate coefficient of variation (CV), and identify the optimal marker to assess TMR mixing quality. TMR was sampled for five consecutive days on a commercial dairy farm, evaluating three diets (high-producing cows, mid-lactation cows, and dry cows). For each diet, three daily feedings were assessed, and in each feeding ten incremental samples were collected along the feed bunk immediately after feed delivery and before animal access. Ten equidistant sampling points were established. The marker was defined as the nutrient showing the highest residual variance among sampling points, with diet as a fixed effect and day and feeding as random effects. Residual variance proportions were: Potassium (99.1%), Crude protein (CP; 96.0%), Ether extract (95.7%), Zinc (94.3%), Phosphorus (93.0%), Neutral Detergent Fiber (NDF; 92.8%), Magnesium (91.3%), Manganese (81.9%), Iron (79.3%), Calcium (77.9%), and Sodium (77.4%). Magnesium was the most consistent marker (CV = 9.5%), while NDF and CP (CV = 9.04% and 9.27%) were also effective indicators of TMR mixing quality. These markers provide a practical approach to monitor TMR mixing quality and improve feed consistency in dairy production systems.

1. Introduction

The confinement of dairy cows fed total mixed ration (TMR) diets represents the predominant production system on commercial dairy farms (Schingoethe, 2017). According to the United States Department of Agriculture (USDA, 2014), approximately 90% of large dairy herds (>500 cows/herd) use TMR as a feeding strategy, as it allows greater control over the diet consumed by cows and facilitates the formulation of balanced rations tailored to different animal categories (Schingoethe, 2017). Under ideal conditions, adequate TMR mixing quality ensures that each portion of the ration delivers the nutrient composition as formulated, thereby supporting consistent nutrient supply to the rumen and optimizing the synchronization between energy and protein for efficient feed utilization (Leonardi & Armentano, 2003; Sova et al., 2014).

The TMR consists of ingredients with diverse physical characteristics and particle sizes, including corn silage, finely ground corn, high-moisture corn, soybean meal, soybean hulls, cottonseed meal, and pellets (Karunanayaka et al., 2022). Proper mixing is essential to ensure random dispersion of these components throughout the ration (Wang et al., 2016), particularly with the increasing inclusion of low-rate ingredients such as vitamins, minerals, amino acids, and other additives, which further heightens the importance of mixing efficiency (Alyami et al., 2017). The accuracy and consistency of the offered diet ultimately determine the composition of the diet consumed by the animals, directly influencing nutrient intake, performance, and feed efficiency.

Consequently, effective nutritional management depends on minimizing discrepancies between formulated and consumed diets, reducing nutrient waste and improving economic returns (Bach, 2024). Poorly mixed diets often require wider safety margins for critical nutrients, leading to reduced compliance with nutritional requirements and suboptimal animal performance. With increasing feed costs, strict monitoring of nutritional balance, commonly referred to as precision feeding, has become increasingly important (James & Cox, 2008). Efficiency is measured by variation in composition between different samples along the feeding line, compared to an ideal mixture (Weiss & St-Pierre, 2024).

The Pearson coefficient of variation (CV) is a common indicator of relative dispersion, used to compare different distributions. Quality control methods in feed management assess mixing effectiveness by checking the uniformity of particle dispersion within a mixing cycle and consistency between consecutive cycles (Cleary & Sinnott, 2008). Efficient processes exhibit a minimum CV within each cycle (homogeneity) and a consistent CV across cycles (uniformity) for all mixed ingredients (Cholette et al., 1959).

Li and Toor (1986) suggested that the evaluation of the mixing process can be conducted by measuring chemical component levels (nutrients or others) in different samples. The use of markers, either internal or external, should ensure statistical homogeneity of variances (homoscedasticity) among ingredient dispersions, considering their physical characteristics (Weiss & St-Pierre, 2024). In feed mills producing non-ruminant feeds, nutrients like manganese are used as quality markers for mixing, with a recommended CV <10% to indicate effective mixing (Herrman & Behnke, 1994; Johnston & Southern, 2000). However, unlike dry concentrates, TMR diets have higher moisture and a complex forage matrix that may affect nutrient dispersion and marker stability, potentially limiting the applicability of feed mill markers under TMR conditions. Despite this, such markers have not yet beeefon established for TMR diets for ruminants, and a methodology is needed to determine the most suitable marker(s). We hypothesized that the quality of TMR mixture for ruminants can be monitored using nutritional markers. The objectives of this study were to validate a methodology, define the appropriate CV, and identify the best nutritional markers to assess the mixing quality of TMR diets for dairy cows.

2. Material and methods

No approval from an Ethics Committee was necessary for this study, as it was conducted exclusively through the analysis of animal diets in the feed bunk at a dairy farm located in Lagoa Formosa, Minas Gerais, Brazil (18°46′ S, 46°24′ W), with cows kept under commercial conditions without interference in their routines. Additionally, the farm staff were not instructed to alter or modify their daily management practices. Performance data were not collected, as this study did not aim to evaluate the effect of the diet mixing quality on animal performance.

2.1. Experimental procedures

To evaluate the mixing quality, TMR samples were collected over five consecutive days at a commercial dairy farm. The study included three types of diets: high-producing cows, mid-lactation cows and dry cows. The main difference between the diets was the proportion of forage to concentrate, ensuring that the mixing quality test covered different diet categories. Each day, three feedings were conducted for each of the three diets.

Samples were collected immediately after the TMR was distributed along the feed bunk and before the cows could access it to avoid interference from sorting. For each batch (dry-cow, high-production, and mid-lactation cows), 10 samples of 500 g each were collected from predefined points along the feed bunk (incremental sampling procedure), with distances set to ensure sample representativeness. For each feeding, 10 samples were collected along the feed bunk, resulting in a total of 450 samples throughout the study (3 diets × 5 d × 3 feedings per day × 10 samples per feeding). Each sample was analyzed individually.

Additionally, each collected sample was analyzed for 11 potential markers present in the three TMR-based diets for lactating cows. The markers evaluated include crude protein (CP), neutral detergent fiber (NDF), ether extract (EE), calcium (Ca), phosphorus (P), potassium (K), magnesium (Mg), sodium (Na), zinc (Zn), manganese (Mn), and iron (Fe). The added sources of these minerals in the diet were limestone for Ca, dicalcium phosphate for P, magnesium sulfate for Mg, common salt for Na, zinc sulfate for Zn, ferrous sulfate for Fe and manganese sulfate for Mn. These markers were selected because CP, NDF, and EE are key nutritional parameters routinely evaluated in dairy diets, and the selected macro- and microminerals are among the most commonly supplemented minerals in lactating cow rations. In addition, all markers are widely available for analysis at relatively low cost, supporting their practical use for evaluating TMR mixing quality.

2.2. Equipment

A horizontal mixer wagon of three axles was used, with a capacity of 1800 kg (4 m³), and equipped with section blades for long fiber, 540 rpm, a one reduction on the main axle, and a side belt distribution. The dosing system used the mixer wagon's scale, with a precision of 3 kg in header/wheel type load cells, with a maximum capacity of 2000 kg, for ingredients ranging from 100 kg to 1000 kg.

Ingredients weighing <100 kg were pre-weighed on an electronic scale with a precision of 0.5 kg in single point load cells, with a maximum capacity of 100 kg. Low inclusion ingredients were weighed on precision scales with a minimum capacity of 0.1 g and a maximum capacity of 200 g, with a precision of 0.01 g. The addition of corn silage was carried out using a loader, while the other ingredients were added manually.

2.3. Feeding management

The diets were prepared and distributed three times a day: at 0600 h, 1000 h, and 1500 h, following the sequence of adding the driest to the wettest ingredient, as described in Table 2. The ingredients were carefully weighed to ensure the accuracy of the diet provided and manually added directly into the mixer wagon, with the axles rotating. The total mixing time was 5 min after the last ingredient inclusion, according to the manufacturer's recommendations for the wagon.

Table 2.

Composition and sequence of ingredient addition to the mixer wagon for the evaluated diets (as-fed basis).

Ingredient, kg	Sequence	High production	Mid-lactation	Dry cow
Concentrate	1	11	8.5	7
Soybean hulls	2	2	1	1
Soybean meal 46%	3	1	1	1
Cottonseed	4	2	2.5	2
Tifton hay	5	0.4	0.4	0.4
Corn silage	6	30	27	30
Water	7	4.5	4.5	3
Inclusion animal/day (kg)		50.9	44.9	44.4

Feed distribution was carried out in a feed bunk line, with the conveyor belt's output flow set at 2000 kg/h, varying the distribution time and the length of the covered feed bunk according to the amount of feed provided for each batch of animals, as it was regularly done at the dairy farm.

Diet 1 was formulated for the high-producing lactating cow group, with a forage-to-concentrate ratio of 65:35, targeting cows with body weight of 650 kg and milk production of 38 kg/d. Diet 2 was designed for the mid-lactation lactating cow group, with a forage-to-concentrate ratio of 68:32, targeting cows with body weight of 650 kg and milk production of 30 kg/d. Diet 3 was formulated for close-up cows, with a forage-to-concentrate ratio of 74:26, targeting cows with body weight of 650 kg, as presented in Table 1.

Table 1.

Ingredients and composition of the formulated total mixed rations (TMR), expressed on a dry matter (DM) basis.

Ingredients, kg	High production	Mid-lactation	Dry cow
Corn silage	10.10	9.09	10.10
Concentrate feed	10.30	7.96	6.55
Soy hulls	0.56	0.43	0.36
Finely ground corn	5.48	4.24	3.48
Soybean meal	0.90	0.70	0.57
DDGS	1.56	1.20	0.99
Urea/corn meal mix	0.22	0.17	0.14
Soypass	1.00	0.77	0.64
Premix	0.57	0.44	0.36
Cottonseed	1.83	2.29	1.83
Soybean hulls	1.78	0.89	0.89
Soybean meal 46%	0.95	0.95	0.95
Tifton hay	0.35	0.35	0.35
Nutrient composition
Dry matter, %	50.34	48.71	47.76
Starch, %	20.03	18.72	17.09
Crude protein, %	16.98	16.94	15.78
RDP, % of CP	62.10	61.92	62.34
RUP, % of CP	37.89	38.06	37.65
Crude fat, %	3.90	4.39	3.92
NDF, %	33.70	33.99	35.38
Macrominerals, %
Ca	0.64	0.60	0.56
P	0.40	0.41	0.39
Mg	0.26	0.27	0.26
K	1.23	1.23	1.23
Na	0.21	0.19	0.17
Trace Minerals, mg/kg
Mn	69.07	65.48	60.92
Zn	84.34	79.28	71.9
Fe	184.94	182.59	185.49

2.4. Laboratory analysis

The chemical analyses were performed to quantify the concentrations of each potential marker candidate in the TMR, as well as to measure the chemical composition of the diets. All samples were prepared according to the M-004/2 method (Detmann et al. 2025). The concentrations of Ca, Mg, Zn, Mn, and Fe were determined using atomic absorption spectrophotometry. Na and K were analyzed using an atomic emission spectrophotometer. Total inorganic phosphorus was measured using a UV/Visible spectrophotometer, following the guidelines of Method M-008/2 (Detmann et al. 2025). The analyses of CP, EE, and NDF were conducted according to Methods N-001/2, G-005/2, and F-002/2, respectively (Detmann et al. 2025).

2.5. Statistical analysis and marker selection for TMR mixing quality

The marker was determined based on the variance of each nutrient along the feed bunk, considering the effect of diet on the variance. The model was analyzed with diet as a fixed effect, and day (nested within diet) and feed bunk section (interaction of day × diet) as random effects. Thus, nutrients with higher variance along the feed bunk were identified as having greater potential to be considered as markers. Additionally, the nutrient CV within diet was evaluated as a support decision tool. We used the proc GLIMMIX of SAS 9.4 (SAS Inc.) with the following model:

$$Y_{ijk}=D_i+P{\left(D\right)}_{ij}+F{\left(D\times P\right)}_{ijk}+e_{ijk}$$

Where:

$Y_{ijk}$ is the nutrient response variable,

$D_i$ is the fixed effect of diet,

$P{\left(D\right)}_{ij}$ is the random effect of period (day) nested within diet,

$F{\left(D\times P\right)}_{ijk}$ is the random effect of feeding time nested within the interaction between diet and period,

$e_{ijk}$ is the general random error with mean of zero and variance (σ²) that is expected to represent the variation between measurements within the same feed bunk.

Therefore, using this model, the nutrient with the greater proportion of the random variance should be the one that captures better the fluctuation in that nutrient composition along the feed bunk.

The mixing quality of TMR was assessed using the proportion of the total residual variance as random variance, which indicates distribution uniformity at collection points, and the CV, which measures data dispersion relative to the mean, with a lower CV indicating greater consistency.

To identify the most suitable component as a mixing quality marker, our research group established minimum criteria based on a residual variance above 90% and the lowest CV among the evaluated components (ideally below 10%) (Herrman & Behnke, 1994; Johnston & Southern, 2000). A high proportion of the total residual variance as random variance indicates that most of the variability observed in the samples is associated with distribution along the feed bunk, which is desirable for a mixing marker, as it means that at least 90% of the variation occurs due to ingredient dispersion at the collection point.

Thus, we aimed for the largest proportion of residual variance to be attributed to random distribution variability rather than systematic factors. Additionally, the smaller the contribution of variance associated with the effects of day and feeding, the better, as this indicates that daily variation and differences between feedings are minimal, reflecting greater consistency in feed management. The analyzed components (Zn, Mn, Fe, Ca, Mg, K, NDF, CP, and EE) were grouped for result analysis. The subgroups were defined as trace minerals (Zn, Mn, Fe), macrominerals (Ca, Mg, K, Na), and macrocomponents (NDF, CP, and EE). This categorization enabled a detailed evaluation of potential markers for assessing TMR mixing quality.

3. Results

3.1. Variance and coefficient of variation as indicators of mixing quality in dairy cow diets

3.1.1. Trace minerals

Zinc showed a high proportion of the residual error variance (94.3%), indicating that most of the Zn variability was attributed to distribution at the collection point. A portion of the error (5.5%) was related to day-to-day variation. This was the only variable that exceeded the established criterion of 90% residual variance. However, the CV for Zn was 26.6%, reflecting variability in distribution (Table 3; Fig. 1).

Table 3.

Proportion of variance (± SEM) and coefficient of variation of trace minerals as markers of mixing quality in dairy cow diets.

Item	Zn		Mn		Fe
Variance	Var1	Prop2,%	Var1	Prop2,%	Var1	Prop2,%
Day (Diet)	28.06±19.34	5.5	34.16±16.99	16.4	1435.83±2697.71	2.9
Feeding (Diet × Day)	0.95±13.45	0.2	3.64±5.43	1.7	8975.79±3453.97	17.9
Residual error	481.5 ±34.12	94.3	170.95±12.06	81.9	39,840.00±2846.15	79.3
Total	510.5		208.74		50,251.62
P-value Diet (Fixed)	0.745		0.180		0.003
Mean, mg/kg	82.6 ±2.99		61.9 ±2.87		529.3 ±34.02
CV3, %	26.6		21.1		37.7
Lower, mg/kg	60.7		48.8		329.7
Upper, mg/kg	104.5		74.9		728.9

¹Variance;.

²Proportion;.

³Coefficient of Variation.

Fig. 1.

Distribution of the proportion of Zn (A), Mn (B), and Fe (C) in the total mixed ration (TMR), expressed in mg/kg. Bars represent the observed frequency, while the red line indicates the fitted normal distribution. The dashed lines mark the lower and upper limits, highlighted by the shaded blue area.

Manganese showed a medium-to-high proportion of the residual error variance (81.9%). Although it indicates a relevant contribution at the collection point, this value did not meet the established criterion of 90%. A proportion of the error (16.4%) was attributed to day-to-day variation, further emphasizing the variability of Mn over time (Table 3).

Iron showed a proportion of the residual error variance of 79.3%, the lowest percentage among the trace minerals analyzed. Unlike the other trace minerals, Fe had a higher variation between feedings (17.9%). Additionally, the CV for Fe was the highest, reaching 37.7%, indicating high dispersion and significant variability in the data (Table 3).

3.1.2. Macrominerals

Calcium showed a residual error variance proportion of 77.9%, which is below the established 90% criterion. The CV was 17.3%. Additionally, 17.6% of the dietary variation in Ca was attributed to day-to-day variation (Table 4; Fig. 2).

Table 4.

Proportion of residual variance (± SEM) and coefficient of variation of macrominerals as markers of mixing quality in dairy cow diets.

Item	Ca		Mg		P		K		Na
Variance	Var1	Prop2	Var1	Prop2	Var1	Prop2	Var1	Prop2	Var1	Prop2
Day (Diet)	0.0029±0.0015	17.6	0.00003±0.00002	5.2	0.00003±0.00009	1.1	0.0003±0.0008	0.6	0.0009±0.0004	22.6
Feeding (Diet × Day)	0.0007±0.0005	4.5	0.00002±0.00002	3.6	0.00017±0.00012	5.9	0.0002±0.0015	0.3	0.0000±0.0001	0.0
Residual error	0.0130±0.0009	77.9	0.00046±0.00003	91.3	0.00273±0.00019	93.0	0.0551±0.0039	99.1	0.0031±0.0002	77.4
Total	0.017		0.001		0.003		0.056		0.004
P-value Diet (Fixed)	0.364		0.093		0.031		0.299		0.927
Mean, mg/kg	0.66±0.027		0.23±0.003		0.28±0.006		1.12±0.018		0.19±0.014
CV3, %	17.3		9.5		18.6		20.9		29.6
Lower, mg/kg	0.55		0.20		0.23		0.89		0.13
Upper, mg/kg	0.78		0.25		0.33		1.36		0.24

¹Variance;.

²Proportion (%);.

³Coefficient of Variation.

Fig. 2.

Distribution of the proportion of Ca (A), P (B), Mg (C), sodium (D), and K (E) in the total mixed ration (TMR), expressed as a percentage of dry matter intake (% DMI). Bars represent the observed frequency, while the red line indicates the fitted normal distribution. The dashed lines mark the lower and upper limits, highlighted by the shaded blue area.

Magnesium showed a proportion of the residual error variance of 91.3%, indicating good uniformity in distribution. Additionally, it had the lowest CV among the macrominerals, at 9.5%, demonstrating high consistency in the data (Table 4).

Phosphorus showed 93.0% of residual variance, with a CV of 18.6% and feeding variation of 5.9%, the highest among macrominerals. Potassium showed 99.1% of residual variance and a CV of 20.9%, indicating strong consistency. Furthermore, K showed minimal variation, with 0.6% attributed to day-to-day variation and 0.3% to feeding variation. Sodium accounted for 77.4% of the residual variance, the lowest among macrominerals. It also had the highest CV (29.6%) and day-to-day variation (22.6%), indicating high dispersion (Table 4).

3.1.3. Macrocomponents

Neutral detergent fiber showed a residual error variance proportion of 92.8%, indicating excellent uniformity in distribution at the collection point. The CV was 9.0%, day-to-day variation was 6.5%, and feeding variation was 0.7%. Crude protein presented a residual error variance proportion of 96.0%, the highest among the macrocomponents and the CV was 8.7%, with 1.7% day-to-day variation and 2.3% feeding variation. Ether extract exhibited a residual error variance proportion of 95.7%, the second-highest among the macrocomponents and the CV was 12.0%, the highest among the macrocomponents (Table 5; Fig. 3).

Table 5.

Proportion of residual variance (± SEM) and coefficient of variation of macrocomponents (ether extract, EE; crude protein, CP; and neutral detergent fiber, NDF) as markers of mixing quality in dairy cow diets.

Item	NDF		CP		EE
Variance	Var1	Prop2,%	Var1	Prop2,%	Var1	Prop2,%
Day (Diet)	0.86±0.54	6.5	0.045±0.069	1.7	0.0075±0.0059	3.4
Feeding (Diet × Day)	0.09±0.34	0.7	0.062±0.089	2.3	0.0021±0.0053	1.0
Residual error	12.21±0.88	92.8	2.558±0.183	96.0	0.2132±0.0151	95.7
Total	13.16		2.665		0.223
P-value Diet (Fixed)	0.291		0.032		0.21
Mean, mg/kg	38.67±0.51		18.35±0.175		3.84±0.053
CV3, %	9.0		8.7		12.0
Lower, mg/kg	35.17		15.72		3.37
Upper, mg/kg	42.16		18.93		4.30

¹Variance;.

²Proportion;.

³Coefficient of Variation.

Fig. 3.

Distribution of the proportion of neutral detergent fiber (NDF) (A), crude protein (CP) (B), and ether extract (EE) (C) in the total mixed ration (TMR), expressed as a percentage of dry matter intake (% DMI). Bars represent the observed frequency, while the red line indicates the fitted normal distribution. The dashed lines mark the lower and upper limits, highlighted by the shaded blue area.

4. Discussion

Precise feed formulation and maximizing the nutritional value of ingredients are crucial for animal performance and the economic viability of production systems (Bach, 2024; Cavallini et al., 2025). Strict control of critical points throughout the entire TMR preparation process, especially during the mixing stage, is essential to ensure ingredient homogeneity (Cavallini et al., 2018).

Most studies focus on the ideal CV for monogastric diets and the quality of concentrate mixes for dairy cows, with limited research on markers for mixing quality in TMR. Markers such as DL-methionine, l-lysine-HCl, and Mn have been used to assess the quality of concentrate mixes (Clark et al., 2007). However, our current results clearly indicate that a CV of 10% is not a reliable metric for assessing TMR homogeneity. Unlike concentrates, TMR exhibits significantly different characteristics, including moisture content and distinct chemical and physical particle properties. Therefore, mixing quality standards established for concentrates should not be directly applied to TMR systems.

Based on the data from Table 4, Zn showed the highest proportion of total residual variance as random variance (94.3%), surpassing Mn and Fe and meeting the minimum threshold of 90% established by our group for assessing mix quality. However, the coefficient of variation for Zn was 26.6%, the highest among the trace minerals, which partially limits its consistency as a marker. From a practical standpoint, such a high CV reduces its reliability for on-farm monitoring, as greater intrinsic variability may mask true mixing inconsistencies, even when the proportion of random residual variance is high. Although Mn is effective as a marker in concentrates (Clark et al., 2007), its use in forage-based diets proved limited, with a proportion of the residual error variance of 81.9%. This difference can be attributed to variations in DM content and pH between concentrates and TMR. While Mn is more soluble in acidic environments, such as those found in conserved forages, its application as a marker in TMR is limited by chemical factors affecting its availability and distribution in the mix (Hem, 1963). In acidic pH, Mn is mainly found as Mn²⁺, which is highly soluble, as in manganese sulfate (MnSO₄), facilitating its dispersion (Härdter et al., 2004). The pH of TMR, typically between 3.8 and 4.5 (Meenongyai et al., 2017), favors this solubility, theoretically making Mn a good marker. However, the effectiveness of Mn decreases due to its tendency to form insoluble complexes in conserved forage diets (López-Rayo et al., 2014). During the mixing process, organic acids such as lactic and acetic acid can bind to Mn, forming insoluble or poorly soluble organometallic compounds, reducing its availability as a marker (Asha et al., 2017).

Additionally, the fiber content of forage sources, such as cellulose and hemicellulose, promotes Mn precipitate formation, making the nutrient less available and hindering its uniform dispersion in the feed (Chandraghatgi, 2003). The high moisture content in TMR also alters Mn solubility, favoring the formation of insoluble compounds such as manganese oxides (MnO₂ and Mn₂O₃), which do not disperse well in the mix (Grangeon et al., 2020), compromising its role as a marker. Although Mn had a lower CV than Zn (21.1% vs. 26.6%), indicating more homogeneous dispersion, its lower proportion of the total residual variance as random variance and interaction with silage compounds make its distribution inconsistent in the feed bunk lane. Thus, Zn and Mn cannot be recommended as primary markers for evaluating TMR mixing quality under forage-based conditions.

Iron proved ineffective as a mix quality marker in TMR due to its chemical instability in diets with variable pH. Iron, provided as ferrous sulfate (FeSO₄), tends to precipitate in acidic environments, such as those with corn silage, forming insoluble compounds that compromise its solubility and uniform distribution (Furcas et al., 2024). Moreover, excess Fe in the diet likely increased the proportion of the total residual variance as random variance (79.3%), impairing mix homogeneity and the effectiveness of Fe as a marker. Therefore, Fe is not suitable as a reliable indicator of TMR homogeneity.

The analysis of macrominerals revealed three potential markers of TMR mixing quality: Mg, P, and K. All showed >90% of the total residual variance attributable to random variation, with Mg standing out for its low CV (9.5%). The high water solubility of magnesium sulfate promotes uniform distribution even under humid conditions (Härdter et al., 2004), and its low hygroscopicity minimizes clumping, thereby enhancing mixture homogeneity (Zhao et al., 2006). Potassium also exhibited a high proportion of the total residual variance attributed to random variation (99.1%), supporting its potential as a marker. However, the practical application of K as a mixing quality marker is more labor-intensive, as potassium is inherently abundant in forage components such as corn silage, hay, and grasses, and its concentration varies substantially with plant physiological stage and fermentation conditions (Erdman et al., 2011). Consequently, the use of K as a reliable marker would require frequent chemical analyses of all forage sources to account for this intrinsic variability, increasing analytical demands compared with minerals supplied primarily through concentrated supplements.

Nevertheless, K exhibited a greater coefficient of variation than Mg, reinforcing the statistical and operational superiority of magnesium as a nutritional marker. When the results for trace and macrominerals are considered collectively, Mg stands out as a promising indicator for evaluating TMR mixing quality. Lower CV values are particularly important, as they reflect minimal intrinsic variability of the nutrient within mixed batches, which is a key requirement for an effective and reliable homogeneity marker. Overall, Mg is recommended as the primary mineral marker for assessing TMR mixing uniformity.

The analysis of Table 5 showed that NDF and CP performed well when combining the proportion of the total residual variance as random variance, with low CVs (92.8% and 96.0%; 9.0% and 8.7%, respectively), and can be considered markers for mixing quality. Both are similar to the results presented by Mg (91.3% for residual variance and CV of 9.5%). Accordingly, NDF and CP represent practical and accessible complementary markers for routine evaluation of TMR mixing quality.

A common approach to evaluating TMR mixing quality is the Penn State Particle Separator (PSPS), which estimates feed particle distribution across sieves. However, the method is susceptible to error and subjectivity, with coefficients of variation reported as high as 30–35% for the upper sieve fraction (Spanghero, 2002). Even under standardized protocols, factors such as shaking intensity, stroke length, operator variation, and sample moisture can substantially affect results (Kononoff et al., 2003), limiting its reliability in commercial settings. Although our study did not evaluate PSPS directly, as the focus was on validating chemical markers, future research should compare these markers with PSPS to assess their relative applicability.

Beyond these methodological considerations, some study-specific limitations should also be acknowledged. The evaluation was conducted on a single well-managed commercial dairy farm using one horizontal mixer over a five-day sampling period, which may limit extrapolation to other production systems, environmental conditions, and seasonal variations. The results may not fully represent other mixer configurations, such as vertical mixers, twin-auger systems, stationary mixers, or alternative loading sequences and mixing times. In addition, only conventional lactation diets were assessed, without inclusion of transition diets with anionic salts, high-fat rations, or diets containing ultra-low inclusion micro-ingredient premixes, which may present different mixing dynamics. Furthermore, no deliberate poor-mixing conditions were imposed; therefore, the sensitivity of the proposed markers to detect severe mixing failures was not directly evaluated. Future studies including multiple farms, diverse mixer types, broader diet formulations, and extended observation periods would further strengthen the external applicability of these findings.

5. Conclusion

Magnesium (Mg) was the most reliable marker for assessing total mixed ration (TMR) mixing quality, combining high residual variance attributed to random variation (>90%) with a low coefficient of variation (CV ≈ 9.5%). Neutral detergent fiber (NDF) and crude protein (CP) showed similarly strong performance (residual variance >92%; CV ≈ 9%) and represent practical, widely available alternatives due to their routine use in feed analysis. In contrast, zinc (Zn) and manganese (Mn) showed greater variability, likely due to chemical interactions in forage-based diets, and iron was unsuitable under acidic TMR conditions. Overall, Mg is recommended as the primary marker for TMR homogeneity, with NDF and CP as effective complementary indicators. A CV threshold of ≤10% is proposed to evaluate mixing quality.

Ethics declaration

No approval from an Ethics Committee was required for this study, as it was conducted exclusively through the analysis of animal diets in the feed bunk. The cows were maintained under standard commercial conditions at a commercial dairy farm in Lagoa Formosa, Minas Gerais, Brazil, with no interference in their routine management.

Farm sta were not instructed to alter or modify their daily management practices, and no experimental treatments were imposed on the animals. In addition, animal performance data were not collected, as the objective of this study was not to evaluate the e ects of diet mixing quality on animal performance.

Marcos Marcondes (representing all authors)

CRediT authorship contribution statement

Erollykens F. Santos: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation. Luciano F. Lago: Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Marina M. Ferreira: Writing – original draft, Visualization, Investigation, Formal analysis, Data curation. Paulo S. Dornelas: Writing – original draft, Visualization, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Fernando A.A. Cidrini: Writing – original draft, Visualization, Supervision, Formal analysis, Data curation. Jardeson S. Pinheiro: Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Julia M.R. Gesteira: Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Data curation. Edenio Detmann: Writing – review & editing, Visualization, Validation, Supervision, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Pedro D.B. Benedeti: Writing – review & editing, Supervision, Project administration, Investigation, Data curation. Marcos I. Marcondes: Writing – review & editing, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

All authors confirm that the research presented in this manuscript was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. No funding bodies, commercial entities, or organizations had any role in the study design; collection, analysis, or interpretation of data; writing of the manuscript; or the decision to submit the manuscript for publication.

Marcos Marcondes (representing all authors)