Experimental diets

Two diets were elaborated with 10% of bean flour (BD) or 10% of alfalfa flour (AD). Agroindustrial Saltillo bean (Phaseolus vulgaris) residue (small and cracked beans) was used, as well as alfalfa (Medicago sativa), cornflour (Zea mays), cane molasses, Mine-Gan, Química Industrial Agropecuaria S.A. de C.V., México), soybean meal flour (47.7% crude protein), sunflower oil and CaCO3. All ingredients were milled and sieved (< 2 mm). Three OC/corn flour percentages ratios were used: 0:55, 3.5:51.5 or 7:48. All other ingredients were kept constant for AD and BD: cane molasses: 5%; soymeal: 5%; CaCO3: 2%; and, glandless cottonseed meal: 23%.

Chemical composition

The chemical composition of raw materials, extrudates, and two commercial diets were determined following (AOAC, 2019) standards.

Extrusion

Samples were processed with a Brabender laboratory simple-screw extruder (Model 2523, 3/4” L/D -25:1 ratio, C. W., Disburg, Germany), with four heating zones. The first three heating zones had a constant T; 90, 100, and 110°C, respectively. The fourth heating zone varied (120, 135, and 150°C), according to experimental design. The screw compression force was 1:1 and the internal diameter of the exit die was 6 mm. Before extrusion, all ingredients were mixed and conditioned to MC of 14, 16, or 18%, following experimental design. After processing, extruded samples were cooled at room T for 4 h and stored in sealed polyurethane bags at 4ºC for further analysis.

Experimental design and data analysis

A factorial design with three independent variables was performed for both diets: OC (X1) [0, 3.5 and 7%], T (X2) [120, 135 and 150°C] and MC (X3) [14, 16 and 18%] before extrusion. Response variables were: expansion index (EI), bulk density (D), water absorption index (WAI), water solubility index (WSI), and hardness (H). Surface response methodology was applied to the experimental data using Design Expert 7.0^® software and results were analyzed by multiple quadratic regressions. Statistical analysis and variance analyzes (ANOVA) for each response were performed using Statistica 7.0^®.

Optimization

Optimization was performed for each OC using a central composite design with three independent variables. Priority responses used for the optimization were: H 25 -50 N, EI 1.0 -1.1, minimum WSI, and maximum D.

Functional Properties

EI and D were measured according to (^{Gujska and Khan, 1990} and ^{Wang et al., 1993}), respectively. WAI and WSI were determined as outlined by (^{Ding et al., 2005}). H was evaluated using a texture profile analyzer model TA-XT2 (Texture Technologies Corp., Scarsdale, NY/Stable MicroSystems, Haslemere, Surrey, UK). In each trial, fifteen samples were sheared using a Warner Brazler blade probe (sensitivity of 1 kgf and 5 cm min^-1) to evaluate breaking strength. ro digestibility and effective degradability

In vitro digestibility (IVD) was calculated using the Daisy II procedure (ANKOM, 2017); inoculum was prepared by diluting ruminal liquid obtained from a 459 kg rumen fistulated cow (criollo race) fed at free grazing, with a buffer solution 1:4 (v/v). Mean comparison was obtained using the Tukey test (P < 0.05). Effective degradability (ED) was calculated following (^{Solanas et al., 2004}) method. Means differences (Duncan, 95%), correlations (Pearson), and t-tests (Student) were performed using Statistica 7.0^®.

Chemical composition

The chemical content of all ingredients (Table 1) is comparable to other reports (^{Reyes-Jáquez et al., 2011}), except for alfalfa’s ash content, which is 2% above reported values, possibly due to a higher presence of fertilizers. Also, alfalfa’s crude protein content was slightly higher than other reports (^{Coblentz and Hoffman, 2009}). THE highest NFE was shown in pinto beans, thus increasing the extrusion capacity of the diets (^{Reyes-Jáquez et al., 2012}).

Functional Properties

All regression coefficients of EI are low (Table 2), possibly because NFE (Table 1) is not high enough to expand, and, process conditions did not yield a significant pressure differential, thus influencing only shaping, not expansion. These results are similar to others (^{Reyes-Jáquez et al., 2011}), where similar diets were extruded at the same T but with higher MC (18 -22%) and without oil. Oil adding in extrudates decreases starch gelatinization due to lower shear stress applied upon the mixture since it acts as a lubricant; it also promotes lipid-starch complexes’ formation, restraining water-starch interaction, resulting in lower EI (^{Liu et al., 2006}). D and H of AD and BD presented significant negative coefficients in all lineal terms (Table 2). High OC percentages generate a less structured and more fragile matrix, because of lipid-starch complexes’ formation (^{Abu-Hardan et al., 2011}); also, high crude fiber content contributes to the fragility of the extruded matrixes (^{Reyes-Jáquez et al., 2012}). High lipid content in extrudates decelerates starch retrogradation and reduces H, adding the fact that high T further degrades starch granules, even leading to dextrinization and creating a less dense and easier to disrupt matrix (^{Rodríguez-Miranda et al., 2012}). High MC generates a higher amount of steam, producing an expanded volume when the mixture is exposed to a pressure differential, resulting in the rupture of matrixes’ cellular walls and creating a more porous product with low D and H (P < 0.05) (^{Reyes-Jáquez et al., 2012}). OC-T interaction presented a significant positive effect (P < 0.05) on D and H, due high concentrations of oil generate more lipid-starch complexes, although high T could lead to break such complexes, allowing starch gelatinization and retrogradation (^{De Pilli et al., 2011}). OC-MC interaction had a positive effect (P < 0.05) on H; (^{Hernández-Hernández et al., 2011}) proposed a model of starch and α-lysophosphatidylcholine (LPC) complexes, indicating a strong competition among lipids and water to bond with starch, and inferring that high MC and OC, are likely to bond starch and water, gelatinizing it and obtaining denser and harder matrixes.

WAI is firstly related to the amount of absorbed water by the starch granules after swallowing in water excess and can be used as a gelatinization grade index (González-Valadez et al., 2008); and, secondly, to the proteins hydrophilic balance in the mixture, which changes according to the denaturalization grade of proteins, where the extrusion process changes solubility profiles (^{Serrano et al., 1998}). Table 2 presents WAI regression analysis: OC and T linear terms affected (P < 0.05) on WAI of both diets. Increasing OC decreases WAI due to lower water availability for the starch granule. As for T, at high values, amylose and amylopectin chains form an expandable matrix that translates into higher water retention capacity. WSI is directly related to starch degradation happening inside the extruder (^{Gujska and Khan, 1990}). Table 2 presents WSI regression analysis: MC linear term, T, and MC quadratic terms of AD; and T, MC, and OC linear terms, MC quadratic term, T, and OC interaction term of BD, had a significant effect (P < 0.05) on WSI. BD’s T negative linear term and T and OC interaction indicate that with high T, proteins denaturalization exposes hydrophobic groups located in the interior, contributing to solubility diminish (^{Ikpeme et al., 2010}).

Optimization

Optimal extrusion conditions were obtained at three different oil concentrations for each diet; AD (0% OC): 142°C, 18% MC; EI: 1.04, D: 1042.7 kg/m³, H: 50 N, WAI: 2.67 g/g, and, WSI: 10.5%; AD (3.5% OC): 131°C, 18% MC; EI: 1.04, D: 1021.7 kg/m³, H: 50 N, WAI: 2.67 g/g, and, WSI: 10.9%; AD (7% OC): 120°C, 18% MC; EI: 1.04, D: 1004.6 kg/m³, H: 25 N, WAI: 2.67 g/g, and, WSI: 10.4%; BD (0% OC): 121°C, 14% MC; EI: 1.09, D: 1125 kg/m³, H: 50 N, WAI: 2.34 g/g, and, WSI: 12.2%; BD (3.5% OC): 120°C, 16% MC; EI: 1.07, D: 1121.3 kg/m³, H: 50 N, WAI: 2.26 g/g, and, WSI: 12%; and, BD (7% OC): 142°C, 15% MC; EI: 1.02, D: 1101 kg/m³, H: 50 N, WAI: 2.27 g/g, and WSI: 11.6%. AD with 7% of OC had the lowest H, while this was not observed with BD. On the other hand, low T for BD with high OC was required compared to AD with low OC. BD needed high T to obtain optimal extrusion characteristics, specifically extrudates with 7% of OC compared to diets with 3 or 0% of OC. Optimization results have the required values to be commercialized due their similar characteristics compared to commercial products as shown in earlier publications (^{Reyes-Jáquez et al., 2011}). Chemical analysis was performed on BD optimal treatment with 7% of OC: MC: 10.7±0.22, protein: 17.6±0.63, fat: 10.7±0.11, fiber: 2.5±0.63 and ash: 8.7±0.18 g/100 g.

In vitro digestibility and effective degradability

IVD of 60.7% or higher is a well-accepted digestibility (^{Coblentz and Hoffman, 2009}), placing AD and BD above such values (Table 3). Excluding both commercial diets, all diets had IVD ranging from 85 to 89% without significant differences (P > 0.05) among them. AD containing 0 or 3.5% of OC has a lower (P < 0.05) IVD than both commercial diets. BD with 3.5 and 7% of OC had comparable IVD to both commercial diets, concluding that incorporating sunflower oil in the extruded diets, increases (P < 0.05) IVD of BD. Table 3 shows the potential degradability (PD) and ED of all diets. PD describes the sum of soluble (A) and insoluble (B) fractions, being B, the maximum degradability peak achieved during 120 h. Table 3 showed that all EDs’ values oscillated between 87.4 and 90.4% with no significant difference (P > 0.05), which are acceptable values compared to other reports (Reyes-Jáquez et al., 2011). It can be concluded that OC and anti-nutritional factors presence did not have a significant difference in IVD nor ED since they were inactivated by high T and shear stress during processing. Also, OC is inside permissible ranges for ruminants’ consumption (^{(Byers and Schelling, 1993}). Extrusion T, MC, and OC affected negatively (P < 0.05) D and H of both diets. However, increasing T and OC, and MC and OC, increases (P < 0.05) H. Obtained IVD was 89.1 and 86.4% for BD and AD with 7% of OC, respectively. Obtained optimal extrusion conditions for BD with 7% of OC were 142°C and 15% of MC; while for AD with 7% of OC were 120°C and 18% of MC. ED was 87.5 and 87.4% for BD and AD with 7% of OC, respectively. Results showed that incorporating sunflower oil in the diets, increases (P < 0.05) IVD, thus opening a high potential opportunity of producing high concentrations of CLA at the ruminal level.