2.1. Diets
Dehulled ground faba beans, ground brewers rice, ground beet pulp pellets and corn gluten meal were all purchased from a local mill (Fairview Mills, LLC, Seneca, KS, USA). Four nutritionally complete experimental diets were formulated, in which rice served as the main carbohydrate source in the control. In the DFB diets, proportional quantities of rice and corn gluten meal were replaced at 10, 20, and 30% of the formula (
Table 1). Diets were formulated to have similar protein, fat, vitamins and minerals content based on previous nutrient analyses. All ingredients were purchased in their ground form and mixed in a 136 kg capacity paddle mixer for 5 min. External markers chromic oxide (Cr
2O
3) and titanium dioxide (TiO
2) were included at 0.25% and 0.4% levels in each experimental diet, for later use in a dog feeding study. The poultry fat and digest (dry dog flavor) were added topically to kibbles after drying.
2.2. Analytical Methods
The nutrient analyses of ration and post-extrusion dietary treatments were determined at a commercial analytical laboratory (Midwest Laboratories, Omaha, NE, USA) as composite samples for each treatment. These included: moisture and dry matter (AOAC 930.15), organic matter and ash (AOAC 942.05), crude protein (AOAC 990.03), fat by acid hydrolysis (AOAC 954.02, modified), crude fiber (AOCS Ba 6a-05), and minerals, including: calcium, phosphorus, potassium, magnesium, sodium, sulfur, copper, iron, manganese, and zinc (AOAC 985.01; modified). Short-chain oligosaccharides including sucrose, raffinose, stachyose and verbascose were measured according to an HPLC method described by Smiricky et al. [
7], and biogenic amines were measured at a commercial laboratory (Midwest laboratories; Omaha, NE) according to procedures described by Shalaby [
8], U.S.FDA [
9], and El Aribi et al. [
10].
2.3. Extrusion Processing
The dry expanded pet foods (
Figure 1) were produced using a single screw extruder (model E525, ExtruTech, Inc., Sabetha, KS, USA). The preconditioner (model 16 × 72 DDC, ExtruTech, Inc., Sabetha, KS, USA) was configured with 12 beaters 45° back, 57 beaters in neutral position, in each of two shafts, and run at a speed of 160 rpm (
Figure S1). The extruder had a defined profile (
Figure S2) and barrel temperatures based on a typical commercial pet food configuration. The target in-barrel moisture was approximately 25% wet basis. Fixed input parameters were kept constant throughout all food production and included dry feed rate (237 kg/h), extruder (EX) water (0%), EX steam (0%) and knife speed (1100 rpm). Variable inputs were considered those controlled by the operator and included pre-conditioner (PC) speed, PC water, PC steam, EX screw speed and mass restriction valve (MRV).
Pre-conditioner and extruder parameters were all collected from sensor readouts during food production. Outputs were those parameters indirectly controlled by the input variables, and included PC discharge temperature, EX die temperature and pressure, total mass flow (TMF), wet bulk density, and specific mechanical energy (SME). Wet bulk density was measured off the extruder four times manually during each respective run, by filling a one-liter cup and leveling the kibbles with a metal ruler and weighing on a digital scale. Total mass flow was calculated by adding the total dry feed rate with water and steam injected in the preconditioner and extruder, assuming that 80% of the water coming from the EX steam is lost during flash-off, as kibbles exit the die:
Specific mechanical energy (SME) was calculated according to the equation below [
11]:
where τ is the % torque or motor load, τo is the no load torque, N is the screw speed, Νr is the rated screw speed (425 rpm), Ρr is the rated motor power (112 kW), and m is the total mass flow (kg/s). The ratio between N and Nr was corrected to be less than or equal to 1.
The size of the kibble was controlled through the extruder die opening (5.2 mm) in order to produce a kibble suitable for feeding small breed dogs. After extrusion, the product was dried on perforated trays in a forced air convection oven at approximately 141 °C until kibbles achieved a moisture level below 9%. The coating consisted of poultry fat and dry palatant and was applied to the kibbles in a double ribbon mixer after they were cooled.
Post extrusion, kibble radial expansion was measured using a digital caliper and expressed as sectional expansion index (SEI), which is the ratio of the cross-section area of the kibble to that of the extruder die. All measurements related to kibble characteristics were done by randomly selecting 10 kibbles per time point for each diet (4 replicates per treatment). Kibbles were weighed using a digital analytical balance, the diameter and length were measured using digital calipers, and these values were used to calculate piece density (g/cm3). Textural hardness and toughness of the products was measured using a texture analyzer (model TA-XT2, Texture Technologies Scarsdale, NY, USA) with a 2.54 cm-diameter cylinder probe. Subsamples of ten random kibbles per extrusion time point of each diet were compressed at 50% strain level. Pre-test speed was 2 mm/s, test speed 2 mm/s, and post-test speed was 10 mm/s. Kibble hardness was the highest breaking force necessary for compression at 50%, while toughness was considered the total force used to compress each kibble by 50%.
2.4. Statistical Analysis
The diets were produced in the order of FB0, FB10, FB20, and FB30 without randomization in an effort to maintain control over target bulk density. For each dietary treatment run, sampling was conducted at evenly spaced intervals and considered replicates for the purpose of determining variability and control during the test. Texture data were analyzed on the averages of the ten randomly selected kibbles from each replicate. Least square means of extrusion output responses were estimated by ANOVA using the GLM procedure with the aid of statistical software (SAS, v. 9.4, SAS Institute INC., Cary, NC, USA), using Tukey correction. Contrasts comparing “control vs. treatment”, linear, quadratic and cubic relationships of diets with graded levels of DFB were considered significant at a p < 0.05.