The impact of marker selection, in-line near-infrared spectroscopy (NIR), and feed mix time on the coefficient of variation (mix uniformity), body weight uniformity and broiler growth performance during the starter, grower, and finisher periods

Abstract

Most feed manufacturers in the United States use the same mixing time (and thus mix uniformity) throughout the growing period regardless of age and consumption patterns. However, research evaluating the optimum mixing time requirements and novel analysis methods, such as in-line near-infrared (NIR) spectroscopy, on the coefficient of variation (CV) and growth performance of broilers throughout the production phases is sparse. Two experiments were conducted to determine the effects of marker selection, in-line NIR, and varying mix times on mix uniformity, broiler growth performance, and body weight uniformity from 1 to 42 d of age. Feed was manufactured utilizing a 1,815-kg counterpoise ribbon mixer. In both experiments, feed was mixed for 4.5 min (3 min dry mix and 90 s of wet mix) and 30 s (0 s dry mix and 30 s wet mix) to obtain a standard mix (SM) and an abbreviated mix (AM), respectively. Experiment 1 constituted a 2 × 2 × 4 factorial arrangement of 2 mix times, (4.5 and 0.5 min), 2 batch sizes (908 and 1,815 kg), and 4 methodologies to evaluate mixer performance (sodium chloride, Microtracers (Red#40 and Blue#40), and in-line NIR). In experiment 2, broilers received different mix time combinations: 1) SM from 1 to 42 d, 2) SM from 1 to 28 d and AM from 28 to 42 d, 3) SM from 1 to 14 d and AM from 14 to 42 d, and 4) AM from 1 to 42 d. In both experiments, selecting a single source marker provided a more accurate estimation of mixer CV in SM and AM diets (P < 0.05). In experiment 2, mix time did not influence BW, feed intake (FI), FCR, or individual bird BW CV from 1 to 42 d of age (P > 0.05). These data indicated that mixer CV differed depending on total mix time and methodology used and diets with a reduced mix time may not necessarily influence growth performance and BW uniformity during the starter, grower, and finisher periods of broilers.

INTRODUCTION

Mix uniformity of broiler diets is a quality parameter that can be controlled depending on the time that ingredients reside in the mixer. Previous research has reported that mixing uniformity is indirectly proportional to mixer capacity and mixing time (Wicker and Poole, 1991; McCoy et al., 1994). Typically, an adequate mix time or uniformity is obtained when a defined marker in the diet has a coefficient of variation (CV) <10% (Beumer, 1991). Poultry consume lower amounts of feed compared to swine and cattle; hence, a longer mixing time and increased nutrient homogeneity have been thought to be required to maximize nutrient consumption and optimize their growth performance (Creger, 1957; McCoy et al., 1994; Poholsky et al., 2021). Currently, the total mix time used in broiler diets throughout the growing period is often 3 to 5 min to maintain nutrient homogeneity and compliance with feed manufacturing regulations (Stark and Saensukjaroenphon, 2017). A good quality mixed and uniform diet has been reported to be essential for growth in nursery pigs and young chicks during the starter period (Ensminger et al., 1990; Traylor et al., 1994; Clark et al., 2007). Nevertheless, if mixing time could be reduced, particularly during the grower and finisher periods, it could be possible to improve feed throughput and reduce labor in high-volume feed manufacturing facilities. Moreover, mix times ≤4 min have been reported to have no detrimental impact in growth performance of swine and poultry (Traylor et al., 1994; Ciftci and Ercan, 2003; Paulk et al., 2015). The evaluation of combined standard and short mix times in specific grow-out periods of broilers is crucial to determine adequate mixing time without compromising growth performance.

Currently, the feed manufacturing and poultry industries perform laboratory assays on medications, amino acids, microminerals, and salt (either the sodium or chloride ion) to evaluate mixer performance (Clark et al., 2007; Stark and Saensukjaroenphon, 2017). Although the measurement of the chloride ion concentration from salt is commonly accepted by feed manufacturers and nutritionists as a uniformity test, poultry diets amended with L-lysine HCL (McCoy et al., 1994; Clark et al., 2007) have produced inconsistent results. Similarly, if salt concentration is measured by analyzing for sodium, and the diet contains a buffer such as sodium bicarbonate, calculated mixer CV may be inaccurate. Therefore, the implementation and exploration of alternative and cost-effective analytical techniques are crucial to obtain an appropriate interpretation of mixer CV, monitor feed quality parameters, and improve feed formulation adjustments.

Near-infrared (NIR) spectroscopy is a type of vibrational absorbance spectroscopy that uses the NIR portion of the electromagnetic spectrum to obtain quantitative and qualitative information about the compositional makeup of target analytes. The use of reflective absorbance (reflectance) NIR has been shown to be advantageous for the analysis and evaluation of animal feeds. Previous research has investigated NIR spectroscopy as a nondestructive method for measuring particle size (Williams and Starkey, 1980; Pasikatan et al., 2001; Ely et al., 2008), physical and chemical characteristics of raw materials (Nielsen et al., 2001; Smith et al., 2001; Owens et al., 2009), adulteration of oils in finished animal feeds (Graham et al., 2012), and quality control in feed conveying systems (Nielsen et al., 2001). The use of NIR spectroscopy for in-line mixer performance determination could be an alternative to continuous quantitative monitoring of CV% of selected markers in broiler diets. The evaluation of the interactive effects of mix time, batch size, and methodologies for testing mixer performance are crucial to optimize mix uniformity (as determined by CV) in broiler diets. Therefore, 2 experiments were conducted to determine the effects of marker selection, In-line NIR spectroscopy, and mix time on mixer CV, body weight uniformity and broiler growth performance from 1 to 42 d of age.

MATERIALS AND METHODS

Feed Formulation, Manufacture, and Experimental Design

In both experiments, dietary treatments were formulated with corn and soybean meal (SBM) as the primary ingredients (Tables 1 and 2). Twelve (experiment 1) and 6 (experiment 2) batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix) and 0.5 min (0 min dry mix and 0.5 min wet mix) to obtain a standard (SM) and an abbreviated (AM) mix, respectively. Experiment 1 constituted a 2 × 2 × 4 factorial arrangement of 2 mix times, (4.5 and 0.5 min), 2 batch sizes (908 and 1,815 kg), and 4 methodologies to evaluate mixer performance (sodium chloride, Microtracers (Red #40 and Blue #40), and in-line NIR). In experiment 2, broiler diets were formulated to meet or exceed the NRC suggested minimum nutrient requirements of broilers (NRC, 1994; Table 2). Each pen was randomly assigned to 1 of 4 dietary treatments: 1) SM from 1 to 42 d of age, 2) SM from 1 to 28 d of age and AM from 28 to 42 d of age, 3) SM from 1 to 14 d of age and AM from 14 to 42 d of age, and 4) AM from 1 to 42 d of age represented by 10 replicate pens.

Table 1.

Ingredient and nutrient composition of a broiler starter diet, experiment 1.

Ingredient, %	Starter
Corn	58.94
Soybean meal, 48% crude protein	31.75
Poultry by-product meal	5.00
Poultry oil	1.00
Monocalcium phosphate, 21% P	0.90
Calcium carbonate	0.84
Sodium chloride	0.50
DL-Methionine (99%)1	0.30
Trace mineral premix2	0.20
Choline chloride	0.20
L-Lysine-HCl (78%)	0.15
L-Threonine	0.10
Vitamin premix3	0.05
Selenium, 0.06%4	0.05
Phytase5	0.01
Microtracers (red6 and blue7 #40)	0.01
	100.00
Calculated analysis, % (unless otherwise noted)
AMEn, kcal/kg	3,000
Crude protein	22.83
Digestible Lys	1.37
Calcium	1.00
Available phosphorus	0.50

Abbreviation: AME_n, apparent metabolizable energy.

¹Donated by Evonik Corporation.

²Mineral premix include per kg of diet: Mn (manganese sulfate), 120 mg; Zn (zinc sulfate), 120 mg; Fe (iron sulfate monohydrate), 80 mg; Cu (tri-basic copper chloride), 10 mg; I (ethylenediamine dihydroiodide), 2.5 mg; and Co (cobalt), 1 mg.

³Donated by DSM Nutritional Products North America. Vitamin premix includes per kg of diet: vitamin A (vitamin A acetate), 6,600 IU; vitamin D (cholecalciferol), 1,980 IU; vitamin E (DL-alpha tocopherol acetate), 33 IU; menadione (menadione sodium bisulfate complex), 2 mg; vitamin B12 (cyanocobalamin), 0.02 mg; folacin (folic acid), 1.1 mg: D-pantothenic acid (calcium pantothenate), 11 mg; riboflavin (riboflavin), 6.6 mg; niacin (niacinamide), 55 mg; thiamin (thiamin mononitrate), 2 mg; D-biotin (biotin), 0.13 mg; and pyridoxine (pyridoxine hydrochloride), 4 mg.

⁴Selenium premix provided Se at 0.3 mg/kg of feed.

⁵Quantum Blue 5G (donated by AB Vista Feed Ingredients, Marlborough, UK) provides per kg of diet: 500 FTU/kg of phytase activity.

⁶Microtracer Red #40 Fe marker added at 50 g/ton.

⁷Microtracer Blue #40 Fe marker added at 50 g/ton.

Table 2.

Ingredient and nutrient composition of dietary treatments fed to Ross × Ross 308 male broilers from 1 to 42 d of age, experiment 2.

Ingredient, % “as-fed”	Starter	Grower	Finisher
Corn	58.89	66.28	71.41
Soybean meal, 48% crude protein	31.75	24.87	20.11
Poultry by-product meal	5.00	5.00	5.00
Poultry oil	1.00	0.90	0.98
Monocalcium phosphate, 21% P	0.90	0.65	0.39
Calcium carbonate	0.84	0.75	0.66
Sodium chloride	0.50	0.47	0.45
DL-Methionine (99%)1	0.30	0.25	0.20
Trace mineral premix2	0.20	0.20	0.20
Choline chloride	0.20	0.20	0.20
L-Lysine-HCl (78%)	0.15	0.17	0.16
L-Threonine	0.10	0.09	0.07
Vitamin premix3	0.05	0.05	0.05
Selenium, 0.06%4	0.05	0.05	0.05
Salinomycin sodium5	0.05	0.05	0.05
Phytase6	0.01	0.01	0.01
Microtracers (red7 and blue8 #40)	0.01	0.01	0.01
	100.00	100.00	100.00
Calculated analysis, % (unless otherwise noted)
AMEn, kcal/kg	3,000	3,070	3,130
Crude protein	22.83	20.17	18.28
Digestible Lys	1.37	1.21	1.08
Digestible SAA9	1.01	0.90	0.80
Calcium	1.00	0.90	0.80
Available phosphorus	0.50	0.45	0.40

Abbreviation: AME_n, apparent metabolizable energy.

¹Donated by Evonik Corporation.

²Mineral premix include per kg of diet: Mn (manganese sulfate), 120 mg; Zn (zinc sulfate), 120 mg; Fe (iron sulfate monohydrate), 80 mg; Cu (tri-basic copper chloride), 10 mg; I (ethylenediamine dihydroiodide), 2.5 mg; and Co (cobalt), 1 mg.

³Donated by DSM Nutritional Products North America. Vitamin premix includes per kg of diet: vitamin A (vitamin A acetate), 6,600 IU; vitamin D (cholecalciferol), 1,980 IU; vitamin E (DL-alpha tocopherol acetate), 33 IU; menadione (menadione sodium bisulfate complex), 2 mg; vitamin B12 (cyanocobalamin), 0.02 mg; folacin (folic acid), 1.1 mg: D-pantothenic acid (calcium pantothenate), 11 mg; riboflavin (riboflavin), 6.6 mg; niacin (niacinamide), 55 mg; thiamin (thiamin mononitrate), 2 mg; D-biotin (biotin), 0.13 mg; and pyridoxine (pyridoxine hydrochloride), 4 mg.

⁴Selenium premix provided Se at 0.3 mg/kg of feed.

⁵Bio-Cox 60 provided salinomycin sodium at 60 g/ton of feed.

⁶Quantum Blue 5G (donated by AB Vista Feed Ingredients, Marlborough, UK) provides per kg of diet: 500 FTU/kg of phytase activity.

⁷Microtracer Red #40 Fe marker added at 50 g/ton.

⁸Microtracer Blue #40 Fe marker added at 50 g/ton.

⁹SAA = total sulfur amino acids.

In both experiments, dry ingredients were blended in a double shaft counterpoise ribbon mixer with a 1,815-kg full load capacity, inlet for major and minor ingredients, a top access door for additional ingredients (hand-adds), and discharge gates (Model TRDB126060, Hayes and Stolz, Fort Worth, TX). In experiment 2, during the starter and grower periods (2 batches/period), SM and AM diets were batched as to utilize half capacity of the mixer on a weight basis (908 kg/batch). The SM and AM finisher diets (2 batches) were batched as 908 and 1,815 kg/batch, respectively.

In both experiments, all fat in the experimental diets was in the form of poultry oil (1%). In the SM diets, dry mix time began after all major, minor, and hand-add dry ingredients were added at the instant the mixer began to run. After the completion of the dry mix cycle, the wet mix cycle began and continued while poultry oil was sprayed into the mixer. In the AM diet, the mixer was manually controlled, and the mix time was determined based on the amount of time required to add poultry oil into the batch of feed (0.5 min). The major and minor ingredients (corn, SBM, and poultry by-product meal) were discharged into the mixer from the major batch scale. All microingredients were individually preweighed into a barrel and added to the mixer at the top access door to prevent a complete dry mix cycle (0 min). For both mix times, the mixing time started when the last ingredient was added to the mixer and ended with mixer discharge, hence it does not include discharge time. Discharge time of the major and microscales was approximately 0.6 and 0.5 min, respectively. After mixing, the mixer was stopped, the discharge gate was opened, and the mixed feed was conveyed by a drag conveyor (4.3 m), elevated (3.7) m, and dropped into a folding bulk container (1,815-kg full load capacity). In both experiments, the experimental diets were manufactured (experiment 1) and fed (experiment 2) in mash form to prevent additional mixing through pelleting and conveying feed systems.

Ten feed samples (0.45 kg of each) were collected from the discharge end (mash leg) using a sampling probe (experiment 1) and 10 feed samples (3 kg of each) were collected from the discharge end (bulk box) using an open-top container (5 gal; experiment 2) and then divided into smaller aliquots for laboratory analysis. In both experiments, mash samples were collected at equally spaced time intervals (approximately 0.5 min, 0.25 min, and 0.13 min/sample) according to previously determined mixer discharge time. Mixed feed discharge time was 5 min/1,815 kg, 2.5 min/908 kg, and 1.25 min/454 kg batch size of feed. The AM diet was packaged continuously, and samples were not split to prevent potential further mixing upon discharge, storage, and laboratory analysis. To analyze mix uniformity, representative subsamples of the SM and AM diets were analyzed, and mixer CV was determined with the use of a “marker” or nutrient (Table 2). The inclusion rate of all markers in the dietary treatments was less than or equal to 0.5%.

Whole corn was ground in a hammer mill (Model 1522, Roskamp Champion, Waterloo, IA) equipped with 2.4-mm and 3.2-mm screens to achieve an average particle size of 779 (experiment 1) and 332 µm (experiment 2). All nutrients and markers selected to determine mixer CV were analyzed for particle size analysis: sodium chloride (391 µm), L-lysine-HCl (78%; 523 µm), DL-methionine (99%; 194 µm), phytase (811 µm), Microtracers RedF #40 (203 µm), and Microtracers BlueF #40 (209 µm; Microtracers Inc., San Francisco, CA; Figure 1). Particle size was determined using a 15-sieve stack with rubber balls, bristle sieve cleaners, and with US sieve numbers 4, 6, 8, 12, 16, 20, 30, 40, 50, 70, 100, 140, 200, 270, and pan. A Ro-Tap shaker (Model RX-29 W.S. Tyler's Ro-Tap, Mentor, OH) was used to sift 100 ± 5 g samples for 10 min. Before sifting the sample, 0.5 g of flow agent (Model SSA-58 Gilson's Inc. Sieving Aid, Lewis Center, OH) was added. Geometric mean particle size by mass (D_gw) and the geometric standard deviation of particle diameter by mass (S_gw) were determined using the quantity of material retained on each sieve following the ASABE method S319.4 (ASABE, 2009). In experiment 1, a coarser corn particle size was used to maintain a continuous flow of feed as it was conveyed and detected by the in-line NIR technology. In experiment 2, the particle size of corn was reduced in order to negate beneficial effects in broiler performance when birds are fed a coarser size of corn (e.g., reverse peristalsis, gizzard development). Providing the bird a diet with a finer particle size should also prevent the selection of coarser pieces of corn when consuming a mash diet as the particles are more homogenous to the rest of the diet ingredients.

Figure 1.

The geometric mean diameter by mass (Dgw) and particle size distribution of markers and nutrients before mixing was determined. The inclusion rate of all markers in the dietary treatments was less than or equal to 0.5%. Twelve (experiment 1) and 6 (experiment 2) batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix) and 0.5 min (0 min dry mix and 0.5 min wet mix) to obtain a standard (SM) and an abbreviated (AM) mix, respectively. The experiment constituted a 2 × 2 × 4 factorial arrangement of 2 mix times (4.5 and 0.5 min), 2 batch sizes (908 and 1,815 kg), and 4 methodologies to evaluate mixer performance (sodium chloride, Microtracers (Red #40 and Blue #40), and in-line NIR). All nutrients and markers selected to determine mixer CV were analyzed for particle size analysis: sodium chloride (391 µm), Microtracers Red #40 (203 µm), and Microtracers Blue #40 (209 µm).

In-Line NIR (Experiment 1)

In the case of the NIR equipment, a Matrix-F FT-NIR (Bruker© Optics, Billerica, MA) in line instrument was inserted into a manufacturing stream to collect spectra readings for further analysis. For this study, real-time, inline analysis was accomplished using an emission-based sensor (Q412/A) connected via a flange fitting to a weld-in sight glass adaptor on the side of a conveyor. The source light from the sensor was used to irradiate the product through the sight glass, with the reflected light being transmitted via fiber optic cable to the FT-NIR spectrometer (Matrix-F). Consistent flow of feed across the interior surface of the sight glass, with sufficient packing density and sample depth to optimize the signal intensity of the light absorbed and reflected by the sample, was critical for this measurement. Spectral data were analyzed using Bruker's spectroscopy software package, OPUS. Quantitative results were predicted using partial least squares (PLS), a type of multivariate regression analysis commonly used for NIR calibration development. A total of 465 readings were obtained from the proximate analysis (ash, fat, fiber, moisture, and protein) throughout the manufacturing of feed at 2 mix times (4.5 and 0.5 min) and 2 batch sizes (908 and 1,815 kg). Although proximate analyses were obtained for each batch of feed, only protein values were selected for mixer CV calculations.

All procedures involving live birds were approved by North Carolina State University Institutional Animal Care and Use Committee (PRN 21-393-A; experiment 2).

Husbandry Practices (Experiment 2)

In experiment 2, Ross × Ross 308 male chicks (Aviagen North America, Huntsville, AL) were obtained from a resident broiler breeder flock housed at NCSU at 1 d of age. A total of 640 male broiler chicks were feather-sexed, weighed and randomly distributed among 40 floor pens (new litter; 16 birds/pen; 0.14 m²/bird) in an environmentally controlled room. Individual birds and feed were weighed to determine body weight (BW), feed intake, and feed conversion ratio (FCR) at 1, 14, 28, and 42 d of age. The facility was equipped with exhaust fans, forced-air heaters, cooling pads, and electronic controllers to manage temperature and ventilation. Each pen was 122 cm in width, 187 cm in length, and 91 cm in height and was equipped with nipple drinkers and 1 tube feeder. Feed and water were offered ad libitum throughout the experimental periods. The lighting program consisted of 23L:1D from 1 to 7 d, 21L:3D from 8 to 20 d, and 16L:8D from 21 to 42 d. The room temperature was 35°C at placement, 31.3°C from 2 to 5 d, 29.4°C from 6 to 14 d, and 28.3°C from 15 to 23 d, 26.7°C from 24 to 28 d and 23.9°C from 29 to 42 d.

Measurements

Coefficient of Variation. In both experiments, subsamples of mash feed were analyzed for sodium chloride, Microtracers RedF #40, and Microtracers BlueF #40, and a coefficient of variation (CV) was calculated for each of the mixing times (Table 3). In experiment 2, subsamples were analyzed for L-lysine-HCl (78%), DL-methionine (99%), and phytase (Table 4). The industry standard for CV to obtain a uniform mix is less than 10%. The coefficient of variation is calculated using the following formula:

$$\text{Mean}\left(\mathrm{\mu }\right)=\sum {\mathrm{x}}_{\mathrm{i}}/\mathrm{n}$$

$$\text{Standard}\text{Deviation}\left(\mathrm{s}\right)=\surd \sum {\left({\mathrm{x}}_{\mathrm{i}}-\mathrm{\mu }\right)}^2/\left(\mathrm{n}-1\right)$$

$$\text{Percent}\text{Coefficient}\text{of}\text{Variation}\left(\%\mathrm{C}\mathrm{V}\right)=\left(\mathrm{s}/\mathrm{\mu }\right)*100$$

Table 3.

Main and interaction effects of mix time, batch size, and methodology on mixer coefficient of variation (CV), experiment 1.

Methodology	Mix time, min6 min	Batch size, kg5	CV, %2
Main effects
Chloride ion (as sodium chloride) chloride)			13.99A
Microtracer Red #40 (count)3			12.53A
Microtracer Blue #40 (count)4			13.04A
In-line NIR			1.49B
P value			<0.0001
SEM1			0.89
	4.5		7.50B
	0.5		13.02A
	P value		<0.0001
	SEM		0.63
		908	10.93
		1815	9.60
		P value	0.1432
		SEM	0.63
Interaction effects
Chloride ion (as sodium chloride)	4.5	-	11.38BC
Microtracer Red #40 (count)	4.5	-	8.72C
Microtracer Blue #40 (count)	4.5	-	8.01C
In-line NIR	4.5	-	1.90D
Chloride ion (as sodium chloride)	0.5	-	16.61AB
Microtracer Red #40 (count)	0.5	-	16.35AB
Microtracer Blue #40 (count)	0.5	-	18.06A
In-line NIR	0.5	-	1.07D
P value			0.0010
SEM			1.25

^A–DMeans within a column with different superscripts differ significantly (P ≤ 0.01).

¹SEM = standard error of the mean for mix time, batch size, and methodology effect (n = 3).

²CV = coefficient of variation (CV = (standard deviation/mean) *100). Mash samples (3 kg of each) were collected at equally spaced intervals using a sampling probe (n = 10) and then divided into smaller aliquots for laboratory analysis. Mixed feed discharge time was 5 min/908 kg.

³Microtracer Red #40 Fe marker added at 50 g/ton.

⁴Microtracer Red #40 Fe marker added at 50 g/ton.

⁵The broiler starter diets were batched as to utilize half or full capacity of a double shaft counterpoise ribbon mixer on a weight basis 908 and 1,815 kg/batch, respectively.

⁶In the 4.5 min mix time diets, after the completion of the dry cycle, the wet cycle began and continued while the poultry fat was sprayed into the mixer. In the 0.5 min mix time diets, the mixer was manually controlled, and the amount of time required was based on the addition of poultry fat into the batch of feed (0.5 min). All microingredients were individually preweighed and added at the top access door to prevent a complete mix. Therefore, dry and wet cycles were done simultaneously while the major and minor ingredients discharged. Discharge time of the major and microscales were approximately 0.6 and 0.5 min, respectively.

Table 4.

Mixer coefficient of variation (CV) of nutrients and markers selected of dietary treatments fed to Ross × Ross 308 male broilers from 1 to 42 d of age, experiment 2.

	Mix time (min)2
	0.5			4.5
Marker, CV %1	143	28	42	14	28	42	Method of analysis
DL-Methionine (99%)	4.94	5.18	8.32	5.62	6.15	9.43	AOAC 999.12/Free AA extraction
L-Lysine HCl (78%)	5.19	8.99	8.31	7.55	8.67	5.01	AOAC 999.13
Chloride ion (as sodium chloride)	10.70	3.85	5.92	3.85	6.29	3.80	Quantab Chloride Titrator
Phytase4	19.95	16.18	21.04	28.16	23.50	37.05	ELISA method
Microtracer Red #40 (count)5	12.19	12.48	14.55	7.81	3.43	6.39	Microtracer Rotary Detector (Handcount)
Microtracer Blue #40 (count)6	12.85	7.13	15.68	4.42	4.18	8.03

¹CV = coefficient of variation (CV = (standard deviation/mean) *100). Mash samples (3 kg of each) were collected at equally spaced intervals using an open-top container (5 gal; n = 10) and then divided into smaller aliquots for laboratory analysis. Mixed feed discharge time was 5 min/1,815 kg, 2.5 min/908 kg, and 1.25 min/454 kg batch size of feed.

²Abbreviated mix (AM): batches of feed were mixed for 0.5 min (0 min dry mix and 0.5 min wet mix). All of the ingredients were added as hand-adds. Mixed feed was packaged continuously in a collapsible black container (1,815 kg full capacity) and samples were not split to prevent potential further mixing upon discharge, storage, and laboratory analysis.

Standard mix (SM): batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix).

³During the starter (14 d) and grower periods (28 d), SM and AM diets were batched as to utilize half capacity of the mixer on a weight basis (908 kg/batch). The SM and AM finisher (42 d) diets were batched as 908 and 1,815 kg/batch, respectively.

⁴Quantum Blue 5G (AB Vista Feed Ingredients, Marlborough, UK) provides per kg of diet: 500 FTU/kg of phytase activity.

⁵Microtracer Red #40 Fe marker added at 50 g/ton.

⁶Microtracer Blue #40 Fe marker added at 50 g/ton.

In experiment 2, feed intake and BW by pen were recorded at 1, 14, 28, and 42 d of age. Birds were observed twice daily, mortalities were removed, and their BW was included in the FCR calculation. The incidence of mortality was recorded daily. Individual BW uniformity by pen was expressed as the CV of BW at 14, 28, and 42 d of age.

Conformity Method (Experiment 1). A conformity method tests whether a sample “conforms” to defined reference material within a certain threshold. Conformity models are built from a collection of “reference samples,” which are samples defined as “good.” These samples meet the criteria to essentially be a control sample that would pass Quality Assurance/Quality Control. Therefore, these samples are representative of the final product (a pass). Additionally, natural variation in the sample and process should be considered for the reference spectra. In the present study, a conformity model was built using the data from the 4.5 min total mix time and a batch size of 1,815 kg as reference spectra. Quantitatively, the average and standard deviation of the absorbance values at each wavelength (i) are calculated. The mean values ± the standard deviation determine the “confidence band” for each absorbance value. Subsequent “test” samples are then measured and compared to the average absorbance values of the reference spectra at each wavelength (i). This absolute deviation is then weighted by the corresponding standard deviation (σ) to calculate the “pass/fail” limit threshold, or conformity index (CI):

$$\text{Conformity Index}\left(\text{CI}\right)=\frac{\left(\text{Absorbanc}{\mathrm{e}}_{\text{sample},\mathrm{i}}-\text{Absorbanc}{\mathrm{e}}_{\text{average},\mathrm{i}}\right)}{{\sigma }_{\text{average},\mathrm{i}}}$$

Statistical Analyses

In experiment 1, results were analyzed as a 2 × 2 × 4 factorial (mix time × batch size × methodologies to evaluate mixer performance) complete randomized design. Data were analyzed using the GLM procedure of JMP software (JMP, 2010) with the following mixed-effects model:

$$Yij=\mu +\rho i+\tau j+\epsilon ij$$

where Yij is the observed response on samples; μ is the overall mean; the ρi are identically and independently normally distributed random effects with mean 0 and variance σ2ρ; the τ j are fixed factor level effects corresponding to the jth dietary treatment (treatments 1 –16) such that τ j = 0; and the εij are identically and independently normally distributed random errors with mean 0 and a variance σ2. The mean values among the 16 treatments were compared using the Tukey's honestly significant different procedure with statistical significance considered at P ≤ 0.05.

In experiment 2, a randomized complete block design was employed with pen location as the blocking factor. Each treatment was represented by 10 replicate pens with pen being the experimental unit. Data were analyzed as a 1-way ANOVA using the GLM procedure of JMP software (JMP, 2010) with the following model:

$$Yij=\mu +\mathrm{T}i+\epsilon ij$$

where Yi j is the observed response of the broilers in the pen; μ is the overall mean; Ti is the fixed effect of mixing uniformity treatment; and εij is the residual error when the pen was regarded as an experimental unit, εij N(0, σ²_ε). The mean values among the combined standard and abbreviated mix times fed to broilers were compared using the Tukey's honestly significant different procedure with the significant level at P ≤ 0.05 unless otherwise indicated.

RESULTS

Experiment 1

The main effects of mix times, batch sizes, and mixer performance methodologies on CV are reported in Table 3. The SM diets and samples analyzed with the in-line NIR generated lower mixer CVs compared to AM, and to the Microtracers and sodium chloride methods (P ≤ 0.01). There were no significant differences due to batch size on mixer CV.

The interaction effects of mix times, batch sizes, and mixer performance methodologies on mixer CV are reported in Table 3. There were no significant interactions of mix times, batch sizes, and mixer performance methodologies on mixer CV (P ≥ 0.05). However, interactions between mix times and methodologies observed in this experiment provide a better understanding of how a longer mix time improves nutrient homogeneity as shown by the different markers selected being distributed uniformly across the batch yielding a lower mixer CV (P ≤ 0.01). The highest mixer CV was observed on AM diets and analyzed with Microtracer Blue #40. Mixer CV was significantly decreased for SM diets when analyzed with Microtracers (Red and Blue #40). In contrast, while the in-line NIR generated the lowest mixer CV, there were no significant differences between the 2 mixing times. Moreover, similarities were also observed for sodium chloride mixer CVs regardless of mixing time.

Experiment 2

The mixer CV of tracer and selected markers, as determined by mix uniformity (total mix time), are reported in Table 4. Although the selected markers and tracers were expected to report a mixer CV >10% in the AM diet (0.5 min) compared with a longer operating mix time, conflicting results were observed. Mix time of mash diets did not influence BW, FI, FCR, CV BW, or the incidence of mortality at 14, 28, and 42 d of age (P > 0.05; Tables 5 and 6).

Table 5.

Growth performance of Ross × Ross 308 male broilers fed with diets containing varying mix times from 1 to 42 d of age, experiment 2.

	Days of age
Item	BW, g/bird4			Feed intake, g/bird			FCR, g:g5			Mortality, %7
Dietary treatments1	14	28	42	14	28	42	14	28	42	14	28	42
1) SM (1–42 d)2	493	1,694	3,158	544	2,210	4,830	1.215	1.342	1.548	0.0	0.6	0.6
2) SM (1–28 d)	491	1,681	3,134	536	2,178	4,805	1.205	1.340	1.550	1.2	1.3	0.6
3) SM (1–14 d)	481	1,680	3,193	543	2,194	4,866	1.255	1.343	1.545	1.9	0.0	0.0
4) AM (1–42 d)3	483	1,667	3,169	538	2,199	4,851	1.236	1.358	1.544	0.0	0.6	1.3
SEM6	9	21	34	7	24	64	0.027	0.011	0.009	0.008	0.006	0.008
P value	0.692	0.840	0.678	0.806	0.823	0.917	0.577	0.626	0.965	0.253	0.555	0.722

^a,bMeans within a column with different superscripts differ significantly (P < 0.05).

¹Treatments consisted of diets with different mix times: 1) standard mix (1–42 d), 2) standard mix (1–28 d) and abbreviated mix (28–42 d), 3) standard mix (1–14 d) and abbreviated mix (14–42 d) and 4) abbreviated mix (1–42 d).

²Standard mix: batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix).

³Abbreviated mix: batches of feed were mixed for 0.5 min (0 min dry mix and 0.5 min wet mix). All of the ingredients were added as hand-adds.

⁴BW = body weight.

⁵Feed conversion ratio was corrected for mortality.

⁶SEM = standard error of the means for mix uniformity effect (n = 10).

⁷Mortality values were arcsin transformed.

Table 6.

Body weight uniformity (expressed as CV) of Ross × Ross 308 male broilers body weight fed with diets containing varying mix times from 1 to 42 d of age, experiment 2.

Item	CV, %4
	Days of age
Dietary treatments1	14	28	42
1) SM (1–42 d)2	13	10	12
2) SM (1–28 d)	16	12	14
3) SM (1–14 d)	17	13	14
4) AM (1–42 d)3	16	11	12
SEM5	2	1	1
P value	0.557	0.302	0.569

^a,bMeans within a column with different superscripts differ significantly (P < 0.05).

¹Treatments consisted of diets with different mix uniformities: 1) standard mix (1–42 d), 2) standard mix (1–28 d) and abbreviated mix (28–42 d), 3) uniform mix (1–14 d) and abbreviated mix (14–42 d) and 4) abbreviated mix (1–42 d).

²Standard mix: batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix).

³Abbreviated mix: batches of feed were mixed for 0.5 min (0 min dry mix and 0.5 min wet mix). All of the ingredients were added as hand-adds.

⁴CV = coefficient of variation (CV = (standard deviation/average pen body weight) *100).

⁵SEM = standard error of the means for mix uniformity effect (n = 10).

DISCUSSION

In agreement with the main effects observed for mix time in experiment 1, a longer mix time decreases mixer CV% because there is a uniform blend of the markers analyzed throughout the batch of feed. Although increased mixing time has been reported to increase mix uniformity in broiler diets (Pfost et al., 1974; McEllhiney and Olentine, 1982; Wilcox and Unruh, 1986; Wicker and Poole, 1991; McCoy et al., 1994), batch size or mixer fill may also have an impact in mixer performance. In agreement, Wicker and Poole (1991) reported that increasing mix time is not sufficient to correct mixing issues caused by overfilling of the mixer. In contrast, Martin (2005) concluded that underfilling the mixer could interrupt the flow of ingredients within the mixing chamber which can negatively impact mix uniformity. Therefore, the lack of differences between batch sizes could suggest that, as long as the mixer is not filled or underfilled beyond rated capacity, it may not impact the degree of variation in mixed feeds.

The adequate selection of a marker or nutrient and type of assay performed are critical to monitor mixer performance in broiler diets. However, differences in sampling method (Reese et al., 2017), accuracy, and costs have an impact in the decision-making process of marker selection and methodology used for determining mix uniformity (Clark et al., 2007). Sodium chloride (salt) can be analyzed through chemical or test strip method, the latter being lower-cost and providing faster results (Ciftci and Ercan 2003). Salt concentration may be determined by measuring either the sodium or chloride ion. In experiment 1, a general consistency between the sodium chloride and Microtracer's mixer CVs was observed. In contrast, McCoy et al. (1994) reported a decrease in CV for sodium chloride concentrations compared with Microtacers when mixing time increased in preliminary mixer evaluations. However, the authors reported no statistical differences on mixer CV when experimental diets were mixed at 20 and 80 mixer revolutions and analyzed for salt and sodium using the Quantab and Omnion methods, respectively. Similarly, Ciftci and Ercan (2003) reported that mixer CVs of sodium chloride decreased from 39.89% (0.20 min) to 13.85% (0.59 min) to 7.95% (3.75 min) using the Merckquant chloride test compared to the titration method which reported no statistical differences at 0.59 and 3.75 min total mix time. A later study reported numerical reductions on sodium chloride and Microtracer Red #40 (absorbance method) CVs when diets were mixed for 0.5, 2.5, and 5.0 min but only the Microtracer Red #40 (hand-count method) and Microtracer RF-Blue Lake (qualitative method) reported statistical differences (Clark et al., 2007). The mixer CV of the Microtracer Red #40 and RF-Blue Lake decreased from 21.77 to 10.43% and from 32.49 to 18.64% when mix time increased from 0.5 to 5.0 min, respectively. Moreover, Zawislak et al. (2011) analyzed mineral mixtures with Microtracer Blue #40 and reported a decrease of 12% in CV when mix time increased from 5 to 8 min but no further decreases at 10 min. In both experiments, test strips were used to measure chloride concentration, and both L-lysine HCl and choline chloride were included in the experimental treatments to replicate broiler diets fed commercially. This could have had an impact on the uniformity test and hence the readings and conversion to %NaCl may not be accurate.

In experiment 2, inconsistencies were observed in the results of mixer CV's based on supplemented DL-methionine (99%), L-lysine HCl (78%), and phytase at different mix times and production phase. The use of all Fe markers (Microtracers) resulted in a CV >10% when the total mix time was 0.5 min, except the Microtracer Blue #40 in the grower diet. Rocha et al. (2015) reported that the use of manganese sulfate (MnS), copper chloride (CuCl), zinc sulfate (ZnS), and sodium chloride (NaCl) can produce confounding CVs as a result of other minor and microingredients containing Mn, Cu, Zn, and Cl. In the case of Microtracers, the use of quantitative vs. qualitative methods could have an impact on the interpretation of uniformity in mixed feeds. Although previous research tested up to 11 different markers simultaneously and reported CVs that ranged from 6 to 54% (McCoy et al., 1994; Clark et al., 2007), the CVs observed with sodium chloride and Microtracers (Red and Blue #40) ranged from 8 to 18% which indicates that these methods provided a degree of variation similar to the standards set by the industry. Therefore, efforts should be made to maintain optimal mixing times, select single source ingredients, monitor assay methodologies, and thus mixer performance to ensure nutrient homogeneity is optimized.

The in-line NIR proved to be a practical tool to determine proximate analysis in real-time. However, the mixer CVs generated by the protein values from the proximate analysis were similar at 4.5 and 0.5 min total mix time. NIR reflectance is sensitive to physico-chemical characteristics found in raw materials and cereal grains (Pasikatan et al., 2001). Although experiment 1 was designed to evaluate the effects of methodologies on mixer performance, previous research has demonstrated that particle size and variation in nutrient profile of ingredients have an impact on the reported chemical composition of feed analyzed by NIR equipment (Mel-cion, 1974; Norris and Barnes, 1976; Pedamond, 1977; Williams and Starkey, 1980; Nathier-Dufour et al., 1995). Since the readings of the NIR are affected by the geometry of the particles, the particle size effects on spectra may be considered “noise,” thus the samples collected must be uniformly ground (Wendtlandt and Hecht, 1966; Pasikatan et al., 2001). The particle size of corn used in the present study was 779 µm to maintain a continuous flow of feed as it was conveyed and detected by the sensor head of the probe.

Prior to further discussion of the NIR analysis, it should be noted that particle size, distribution, and other physical characteristics (e.g., shape and density) also impact other methodologies. For example, Creger (1957) reported that the higher density of sodium chloride allowed the particles to “sift” from the carrier (SBM) while the lower density of particles in nitrophenide caused them to “adhere” to the particles of SBM, hence explaining the improved mixing capabilities of sodium chloride with the carrier in a shorter period of time. The authors reported a similar effect when sodium chloride and nitrophenide were mixed with corn ground through a 1/8 inch screen and concluded that the particle size distribution of markers play a role in mix uniformity. The geometric mean diameter of the markers used in the present experiment is shown in Figure 1. Even though both Microtracers had similar particle size (203 and 209 µm), the sodium chloride used in this experiment was slightly coarser (391 µm) and followed a different particle size distribution. This could explain the lower mixer CV% obtained with sodium chloride regardless of mix time.

In the case of NIR technology, the overall particle size of the feed samples analyzed by the in-line NIR were not uniform across all batches. Furthermore, the number of readings obtained with an in-line NIR depends on the amount of feed flowing through the welded window at the conveyor. Mateo-Ortiz et al. (2014) reported that probe location or position and paddle wheel speed have an impact on NIR predictions in a tablet press. A batch size of 1,815 kg allows more time for data collection by the equipment compared to a batch size of 908 kg which flows at a faster speed and may not cover the window uniformly for the probe light beam.

In experiment 1, it is critical to acknowledge the importance of single source markers as a rationale for why the response on mixer CV from the protein values obtained from the experimental diets was quite lower than other methodologies. This is to be expected as the analysis does not distinguish between sources of protein, which was supplied by multiple ingredients included in the experimental diet. However, additional data collected during the experiment indicate that the in-line methodology may indeed be useful if data beyond the simple proximate value.

Previous research has reported that most of the protein analyses are affected by the physical composition of other ingredients in the diet (Williams and Starkey, 1980). In general, the default CI limit is 3; nevertheless, specific spectral regions from observations of a previous model were considered, thus a CI limit of 2.8 was used for this model. A CI limit of 2.8 was ideal because the majority of the test spectra “failed” which is expected for different mix times and batch sizes (4.5 min/908 kg, 0.5 min/908 kg, and 0.5 min/1,815 kg) differing from the reference (4.5 min/1,815 kg). Effectively, the analysis expects to see a uniform sample matching the reference, but in many cases does not. However, out of 63 test samples, 6 “conformed” or passed the test which belong to the test spectra of 1,815 kg with a total mix time of 0.5 min. The comparison of both mix times and batch sizes within the conformity test are reported in Figure 2. To have a better understanding of the response observed, 2 datapoints (#33 and #58) were highlighted in yellow in the graph. Although both samples correspond to a total mix time of 0.5 min, sample #33 (1,815 kg) has a max CI value of 11.43 while sample #58 (908 kg) correspond to a different batch size and reported a max CI value of 3.20. Therefore, considering both batch sizes, a shorter mix time increases the differences observed on CI values from the reference spectra. Moreover, graphic representations of the raw spectra for the 0.5 min total mix time and the reference spectra are reported in Figures 3 and 4, respectively. A closer look to the raw spectra collected by the in-line NIR facilitated a visual representation of the differences observed in the spectra readings between runs.

Figure 2.

Graphic representation of a conformity test for the 4.5 and 0.5 min mix times and 908 and 1,815 kg batch sizes, experiment 1. Twelve batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix) and 0.5 min (0 min dry mix and 0.5 min wet mix) to obtain a standard (SM) and an abbreviated (AM) mix, respectively. The experiment constituted a 2 × 2 × 4 factorial arrangement of 2 mix times (4.5 and 0.5 min), 2 batch sizes (908 and 1,815 kg), and 4 methodologies to evaluate mixer performance (sodium chloride, Microtracers (Red #40 and Blue #40), and in-line NIR). To build the conformity model, only 2 mix time and batch size were considered. A conformity index (CI) of 2.8 was determined (red line). The reference spectra are indicated in green and the test spectra in blue. The yellow points correspond to sample #33 and #58, both with a 0.5 min total mix time but 1,815 and 908 kg batch sizes, respectively.

Figure 3.

Graphic representation of the raw spectra (no preprocessing) for the 0.5 min total mix time, experiment 1. Twelve batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix) and 0.5 min (0 min dry mix and 0.5 min wet mix) to obtain a standard (SM) and an abbreviated (AM) mix, respectively. The experiment constituted a 2 × 2 × 4 factorial arrangement of 2 mix times (4.5 and 0.5 min), 2 batch sizes (908 and 1,815 kg), and 4 methodologies to evaluate mixer performance (sodium chloride, Microtracers (Red #40 and Blue #40), and in-line NIR). The dark blue spectra represent all of the runs for 0.5 min mix time and 908 kg batch size. However, the runs for 0.5 min and 1,815 kg batches were separated: bright blue (run #7), magenta (run #8), and yellow is run (#12).

Figure 4.

Graphic representation of the raw spectra (no preprocessing) for the reference spectra (4.5 min total mix time and 1,815 kg), 4.5 min—908 kg, and 0.5 min—908 and 1,815 kg, experiment 1. Twelve batches of feed were mixed for 4.5 min (3 min dry mix and 1.5 min of wet mix) and 0.5 min (0 min dry mix and 0.5 min wet mix) to obtain a standard (SM) and an abbreviated (AM) mix, respectively. The experiment constituted a 2 × 2 × 4 factorial arrangement of 2 mix times (4.5 and 0.5 min), 2 batch sizes (908 and 1,815 kg), and 4 methodologies to evaluate mixer performance (sodium chloride, Microtracers (Red #40 and Blue #40), and in-line NIR). The reference spectra (4.5 min and 1,815 kg) in green with all of the spectra for the rest of the runs.

The suitability of the CV analysis methods used in broiler diets is influenced by analytical assays, the inclusion rate of feed ingredients, common nutrient sources, and markers or tracer selection, which can result in over- or underestimation of mix uniformity. McCoy et al. (1994) concluded that broilers could be fed diets with a mixer CV of 20% without a negative impact on growth performance. Therefore, the effects of CV methodologies in selected markers or tracers in relation to mixing time require further evaluation to determine a suitable mix uniformity in broiler diets. However, in experiment 2, the authors acknowledge that although not all the mixer CVs results obtained were considered “poor” (>10%), the mix time could not be decreased any further with the batching and mixing equipment available. It is also worth noting that the type of mixers and/or mixer capacities used in previous studies are not necessarily representative of the modern mixers installed in new and/or renovated facilities. Furthermore, even with the mixing equipment used and with higher CV values, there was not a detrimental impact on the animal's growth performance which will be discussed in the following section.

In agreement with experiment 2, McCoy et al. (1994) reported similar effects on broilers fed diets containing different mixer revolutions (mix times) without a negative effect on growth performance during the starter period. In contrast, Groesbeck et al. (2007) reported that average daily gain (ADG) increased from 190 to 280 g and gain:feed from 0.71 to 0.90 as mix time increased from 0 min to 5.5 min at 14 d of age in pigs. Similarly, Traylor et al. (1994) reported an increase in ADG and average daily feed intake (ADFI) on nursery pigs by increasing mix time from 0 to 0.5 min. Even though a minimal increase in mix time improved growth performance, the authors stated that manufacturing diets with a significantly shorter mix time than the standard is not suitable for optimum growth in young pigs. Previous research has reported that mix uniformity has a greater impact on younger chicks and pigs because they consume less feed during the starter period when compared to periods thereafter (Ensminger et al., 1990; Traylor et al., 1994; Rocha et al., 2022). Rocha et al. (2022) reported no detrimental impact on growth performance when feeding diets with a mixer CV of 22.6% to male broilers from 12 to 40 d of age. However, in the present experiment, feeding an AM diet during the starter period did not have a negative impact on growth performance of young chicks.

Similarly, Traylor et al. (1994) reported that mix times of 0, 0.5, 2, and 4 min had no effect on growth performance of finishing pigs. The authors stated that the experimental treatments in their study had additional mixing time through the feed conveying systems which could have increased uniformity of the 0 min mix time. Ciftci and Ercan (2003) reported no differences in BWG, FI, FCR, and mortality on broilers fed diets containing mix times of 0.20, 0.59, and 3.75 min at 42 d of age. These data indicated that, although a detrimental impact in growth performance was expected by decreasing mix time from 4.5 to 0.5 min, broilers tolerated a diet mixed for a shorter time in the late stages of their feeding phases. According to the Food and Drug Administration's (FDA) Good Manufacturing Practices, it is expected that animal feed in compliance with regulations will have a suitable uniformity of nutrients in diets and feed additives (Traylor et al., 1994; Muirhead, 2006), but rules do not specifically identify required mixer CV values. Therefore, the total mix time for broiler diets fed during the grower and finisher periods may be reduced without compromising broiler growth performance.

A recent study reported no differences in CV of 27 d and 33 d BW in finishing pigs fed diets containing mix times of 0, 0.5, 2, and 6 min (Paulk et al., 2015). In addition, the authors reported that increasing mix time from 0 to 6 min reduced the CV of markers added in feed such as salt and chromium by 39 and 36%, respectively. However, these improvements in ingredient variation were not sufficient to impact BW uniformity of finishing pigs. Ciftci and Ercan (2003) reported similar effects on broilers fed diets containing different mix times (0.20, 0.59, and 3.75 min) without negatively affecting the CV of 42 d BW. Previous research has reported that nutrient deficiency, variation in genotype, and environmental conditions inside a broiler house have a significant impact on BW uniformity within a flock of broilers (Al Homidan et al., 1998; Gous, 2018). Gous (2018) concluded that broilers fed diets containing higher inclusion rates of protein and vitamins and cooler house environment results in increased BW uniformity of broiler flocks. In addition, Xu et al. (2015) reported that feeding mash diets decreases BW uniformity compared with crumbled and pelleted diets. In the present study, diets were fed in mash form to reduce additional mix time by feed conveying systems and a selected feed allocation of the packaged AM diets was used to increase nonuniform nutrient intake and uniformity within the flock. However, the renovated conditions of the broiler house and cool weather conditions provided a steady microenvironment and stimulated feed intake of broilers which may have precluded differences in BW uniformity. Even though birds fed the SM diets were expected to outperform those fed AM diets during the starter period, the impact of the AM diets on growth performance was negligible in subsequent phases. These data demonstrated that diets with a total mix time of 0.5 min can be fed to broilers during the grower and finisher periods without compromising growth from 28 to 42 d of age.

CONCLUSIONS AND APPLICATIONS

• 1.The mixer CVs of sodium chloride (Cl ion; experiment 1 and 2) and protein from the proximate analysis of the in-line NIR (experiment 1) provided cofounding results. However, in both experiments, Microtracers (Red and Blue #40) performed as expected to estimate mix uniformity, which emphasizes the importance of using a single source as a marker that will not interfere with other ingredients in the diet.
• 2.In experiment 1, although the markers selected estimated differences on mixer CV performance at 2 mix times and batch sizes, a further exploration of the conformity test and spectra readings with the in-line NIR provided practical representations of the behavior of samples during mixer discharge of finished feed, and suggest that spectral differences rather than differences in analyzed proximate values, may be the key to utilizing in-line NIR for real-time uniformity analysis.
• 3.Broilers were able to consume an AM diet without any adverse effects on BW, BWG, FI, FCR, mortality, and BW uniformity from 1 to 42 d of age.
• 4.Mix time of mash diets did not compromise growth performance and BW uniformity during the grower and finisher periods, which indicates that a shorter mix time may be suitable for broilers in subsequent phases, especially when utilizing modern high-speed mixing equipment. Therefore, future research should evaluate the effects of feeding pelleted diets with reduced mixer CV on growth performance during the grower and finisher periods of broilers.