CIAT Research Online - Accepted Manuscript Optimization of Roba1 extrusion conditions and bean extrudate properties using response surface methodology and multi-response desirability function The International Center for Tropical Agriculture (CIAT) believes that open access contributes to its mission of reducing hunger and poverty, and improving human nutrition in the tropics through research aimed at increasing the eco-efficiency of agriculture. CIAT is committed to creating and sharing knowledge and information openly and globally. We do this through collaborative research as well as through the open sharing of our data, tools, and publications. Citation: Natabirwa, Hedwig, Nakimbugwe, Dorothy, Lung'aho, Mercy, Muyonga, John H. (2018). Optimization of Roba1 extrusion conditions and bean extrudate properties using response surface methodology and multi-response desirability function. LWT - Food Science and Technology, 96, 411–418. Publisher’s DOI: https://doi.org/10.1016/j.lwt.2018.05.040 Access through CIAT Research Online: http://hdl.handle.net/10568/93204 Terms: © 2018. CIAT has provided you with this accepted manuscript in line with CIAT’s open access policy and in accordance with the Publisher’s policy on self-archiving. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. You may re-use or share this manuscript as long as you acknowledge the authors by citing the version of the record listed above. You may not change this manuscript in any way or use it commercially. For more information, please contact CIAT Library at CIAT-Library@cgiar.org. 1 Optimization of Roba1 bean extrusion conditions and extrudate properties using 2 Response surface methodology and multi-response desirability function 3 Hedwig Natabirwaa,b,*, Dorothy Nakimbugwea, Mercy Lungahoc, & John H Muyongaa 4 a School of Food Technology Nutrition & Bioengineering, Makerere University, P.O Box 7062, 5 Kampala, Uganda 6 b National Agricultural Research Laboratories, National Agricultural Research Organization, 7 P.O. Box 7065, Kampala, Uganda 8 c Center for International Tropical Agriculture, P.O. Box 6247, Kampala, Uganda 9 10 *Corresponding 11 University, P.O Box 7062, Kampala, Uganda. E-mail: hedwignorh@yahoo.com author: School of Food Technology Nutrition & Bioengineering, Makerere 12 13 Abstract 14 Effects of extruder die temperature, screw speed and ingredient feed moisture on 15 Roba1 bean extrudate nutritional and physicochemical properties were evaluated by 16 response surface methodology (RSM) and extrusion processing conditions 17 optimized for optimal extrudate attributes by multi-response desirability function. 18 Responses taken were protein content, protein digestibility, polyphenols, phytates, 19 extrudate expansion, bulk density, water absorption and water solubility index, as 20 well as texture.. Feed moisture, die temperature and screw speed significantly (p < 21 0.05) influenced the physicochemical properties of Roba1 extrudates (R2  0.500). 22 Increase in feed moisture at low die temperatures resulted in decrease in extrudate 1 23 expansion ratio (~3.96%) and water solubility (~10%). Increases in expansion, and 24 reduction in bulk density and water absorption index due to increase in screw speed 25 and die temperature were also observed. Predictive desirability optimization 26 generated optimal attributes (expansion ratio, 2.59; bulk density, 1.32; protein 27 digestibility, 81.58%; and hardness, 24.4 N) for snack with desirability index of 0.75. 28 Information from this study can be useful for optimization of bean snack extrusion 29 process and product in the food industry. 30 Key words: Extrusion, common beans, optimization, desirability function, 31 physicochemical properties, RSM 32 INTRODUCTION 33 Common beans (Phaseolus vulgaris L.) are a nutritious food consumed as a staple 34 by a large population throughout the world, especially in developing countries 35 (Anderson et al., 2016). Beans not only provide proteins, but are also rich in 36 vitamins, minerals and fibre (Nyombaire, Siddiq, & Dolan, 2011), and average daily 37 per capita consumption ranges between 0.01 and 0.18 kg/day (Blair, 2013). 38 Recently, new common bean varieties enriched with iron and zinc through 39 conventional breeding were developed with the aim of reducing micronutrient 40 malnutrition among vulnerable populations. Among these is Roba1 containing 41 about 66.7 and 27.6 g/g of Fe and Zn, respectively (Natabirwa, Muyonga, 42 Nakimbugwe, & Lungaho, 2018). Roba1 is a high-yielding and multiple disease- 43 resistant haricot bean variety currently produced in Sub-Saharan African countries 44 (Ethiopia, Tanzania and Uganda), with average yield of about 1.8 tons/ha 45 (Mukankusi, Nkalubo, Katungi, et al, 2015).Biofortified and iron-rich beans in 46 general ontain higher levels of iron (~70 - 96 ug/g) and zinc and (~26 - 35ug/g) than 47 the conventional beans (Bouis & Welch, 2010). The use of biofortified beans as 2 48 major ingredients in processed foods would help to improve the nutritional quality. 49 Bean consumption however has generally remained relatively low due to the lengthy 50 cooking time which translates into high fuel costs (Rocha-Guzman et al., 2008), and 51 monotonous preparation techniques involved. Moreover, common beans have 52 hardly been explored as raw material in the food industry, except canning and to 53 some extent, flour production (Nyombaire et al., 2011; Pedrosa et al., 2015). Even 54 though sprouting, soaking, fermentation and roasting have been explored for 55 nutritional improvement, the processes are cumbersome and can only be used on 56 small-scale (Nkundabombi, Nakimbugwe, & Muyonga, 2015; Rehman, Salariya, & 57 Zafar, 2001; Rocha-Guzman et al., 2008). 58 Extrusion cooking, a high temperature-short time industrial processing technique, 59 has been singled out as the most promising method which can transform food raw 60 material into highly nutritious, palatable and quality food (Ghumman, Kaur, Singh, 61 & Singh, 2016; Siddiq, Kelkar, Harte, Dolan, & Nyombaire, 2013). Extrusion 62 produces a range of food products from cereals and legumes (including flours, 63 snacks, breakfast cereal, etcetera) with distinctive characteristics (Anton, Fulcher, & 64 Arntfield, 2009; Meng, Threinen, Hansen, & Driedger, 2010; Nyombaire et al., 65 2011). The technology offers high efficiency in terms of fuel cost and output, but it 66 is affected by a range of factors including ingredient type and extrusion conditions 67 which determine the properties of extrudates (Ghumman et al., 2016; Steel et al., 68 2012). Notably, any variations in parameters such as extruder barrel temperatures, 69 ingredient moisture, specific mechanical energy and screw speed, affect process 70 variables as well as product quality. 71 Reports have shown that mild extrusion conditions (high moisture content, low 72 residence time, low temperature) improve the nutritional quality of beans, while high 73 extrusion temperatures (>200 °C), low moisture contents (< 15%) and/or improper 3 74 formulation (such as presence of high-reactive sugars) can adversely impair the 75 nutritional quality (Berrios, Ascheri, & Losso, 2012; Siddiq et al., 2013). However, 76 studies on effects of extrusion processing conditions on extrudate properties have 77 majorly been undertaken on conventional beans, other legumes and cereals 78 (Ghumman et al., 2016; Korus, Gumul, & Czechowska, 2007; Nyombaire et al., 79 2011; Rathod & Annapure, 2017; Siddiq et al., 2013). 80 Response Surface methodology (RSM) and multi-response desirability function can 81 be used for optimization of processing conditions through exploration of relationship 82 between several processes and responses (Altan, McCarthy, & Maskan, 2008; 83 Bezerra, Santelli, Oliveira, Villar, & Escaleira, 2008; Jain, Monika; Singh, Chetna; 84 Gupta, Kushboo; Jain, 2014). The RSM approach is important in design, 85 development and formulation of new products, as well as improvement of existing 86 product design (Bezerra et al., 2008). With the desirability function approach, 87 operating conditions that meet the criteria set for optimization and provide the best 88 value of compromise for combined responses, are established (Vera Candioti, De 89 Zan, Cámara, & Goicoechea, 2014). Various product quality characteristics can be 90 optimized together. For the extrusion of biofortified beans to be successful, critical 91 control of processing conditions and a precise study of the variations that occur in 92 product properties are necessary. The objective of this study was to investigate the 93 effects of feed moisture, screw speed and die temperature on physicochemical and 94 nutritional properties of Roba1 bean extrudate and; to determine the best 95 combination of extrusion processing parameters for a desirable snack extrudate. 96 97 98 MATERIALS AND METHODS 2.1 4 Experimental design 99 Roba1 bean extrudates were developed following a Box Behnken design with three 100 independent variables including ingredient feed moisture, extruder die temperature, 101 and screw speed. Levels of each variable were established basing on preliminary 102 trials and works from previous authors (Anton et al., 2009; Berrios et al., 2012). The 103 three levels of process variables were coded as -1, 0 and 1, making the total number 104 of experiments equal to 15 by Box Behnken design. Coded and actual values for 105 process variables are given in Table 1. 106 Table 1 Coded and actual values used in developing experimental data Factors Factor codes Die temperature (°C) Feed moisture (%) Screw speed (Hz) 107 𝑋1 𝑋2 𝑋3 Level codes and actual values -1 0 1 120 135 150 15 17.5 20 35 40 45 108 2.2 109 Newly harvested dry beans of variety Roba1, a plain cream coloured bean enriched 110 with iron and zinc through biofortification, were purchased from farmers in Rakai 111 district, Uganda. The beans were sorted, washed with clean tap water and solar dried 112 at temperatures 30 – 55°C for approximately 20 hours. Dried bean grains were 113 milled using a commercial mill (Model YZMF, Yize, Shuliy Henan, China), to pass 114 through a 1.5 mm sieve. 115 2.3 116 Beans were extruded in a Twin Screw Extruder (Model DP 70-III, Jinan, China) at 117 barrel temperatures 60/100/120°C, 60/110/135 °C and 60/110/150 °C, and feed 118 moisture 15 to 20 % and screw speeds (35, 40 and 45 Hz), as described in Table 1 119 above. The die diameter, screw diameter and length to diameter ratio of extruder 120 were 5 mm, 27 mm and 18:1, respectively. Resultant extrudates were cooled to room 5 Material preparation Extrusion 121 temperature, and milled using a Stainless Steel mill (Model 30B-C, Changzhou, 122 China) to pass through a 1.5 mm pore size sieve. 123 2.4 124 Extrudate analysis Physicochemical properties 125 Extrudate expansion ratio (ER) and bulk density (BD) were determined according to 126 methods described (Natabirwa, Muyonga, Nakimbugwe, & Lungaho, 2017). Water 127 absorption index (WAI) and water solubility index (WSI)were determined using 128 methods described (Natabirwa et al., 2017; Nyombaire et al., 2011). The final 129 pasting viscosity of extruded flour was determined using extrusion profile on a Rapid 130 Viscoanalyzer, Model RVA 4500, (Perten Instruments, Australia) using methods 131 described (Natabirwa et al., 2017). 132 Extrudate texture 133 The texture of cylindrical extrudates (approximately 3 – 4 cm long pieces) was 134 measured using a Stable Microsystems Texture Analyzer (Model TA.XT-Plus 135 42095, UK) by compression with a cylindrical probe of 6 mm diameter (SMS P/6) 136 following methods described (Altan et al., 2008) with modifications. Hardness in 137 newtons (N) was determined by measuring the maximum force required to break the 138 extruded samples (~ 40 mm long), while crunchiness was determined as the average 139 area under the force-deformation curve. The test speed was 2mm/s and the 140 penetration distance was 5mm, with a trigger force 0.049 N. Post-test speed was 141 10.00 mm/sec. The return distance of the probe was kept at 20 mm. A force time 142 curve was recorded and analyzed by Texture Exponent 32 Software programme. Ten 143 (10) measurements were performed on each sample and averaged. 144 Protein digestibility 6 145 A multi-enzyme in vitro technique (Hsu, Vavak, Satterlee, & Miller, 1977; Krupa- 146 Kozak & Soral-Śmietana, 2010) was applied for determination of protein 147 digestibility since it could avoid under-predicting the digestibility of proteins. The 148 digestibility was calculated using the regression equation (eq. 1) (Hsu et al, 1977). 149 𝑃𝑟𝑜𝑡𝑒𝑖𝑛 𝑑𝑖𝑔𝑒𝑠𝑡𝑖𝑏𝑖𝑙𝑖𝑡𝑦 (%) = 201.464 − 18.103 × 𝐻 (1) 150 where, H is the pH value of the sample suspension after 10 minutes digestion with 151 the multi-enzyme solution. 152 Total polyphenol content 153 Total polyphenol content was determined using the Folin-Ciocalteau reagent, as 154 described (Makkar, 2000; Natukunda, Muyonga, & Mukisa, 2016), with 155 modifications. Briefly, 0.20 g of finely ground bean flour was measured into a 50 ml 156 polypropyrene tube and extracted twice using 5 ml of methanol:water (50:50, v/v) 157 with ultrasonication (20 min). The extraction solution was centrifuged at 3000 x g, 158 for 10 min.), and the supernatant (extract) was stored at 4°C in a refrigerator until 159 time for use. Total polyphenols in the extract were then determined as described 160 (Natukunda et al., 2016). A standard curve was prepared with gallic acid at 161 concentrations 0.00, 0.02, 0.04, 0.06, 0.08 and 0.10 mg/ml. Final results were 162 expressed as gallic acid equivalents (mg GAE/100 g of bean flour). 163 Phytate determination 164 Phytate content was determined according to the method described (Gao et al., 2007) 165 with modifications. Briefly, 0.5 g of raw and extruded bean flour (ground to pass 166 through a 1.0 mm screen) was weighed into clean 15 mL polypropyrene tubes, and 167 10 mL of 2.4% HCl extraction solution was added. The tubes were shaken at 220 168 rpm for 16 h in an Orbital Incubator/Shaker (Model, Stuart S1600C, Wagtech) and 169 centrifuged at 1000 x g for 10 min. at 10 °C. The extract was collected into a new 7 170 set of falcon tubes containing 1 g NaCl each. The contents were vortexed for approx. 171 1 min. to dissolve the salt and allowed to settle at 4 °C for 60 min. The mixtures were 172 centrifuged again at 1000 x g for 10 minutes, and clear supernatants were collected 173 for colour development. An aliquot of clear supernatant (1 mL) was diluted 25 times 174 in a 50-mL polypropyrene tube with distilled deionized water. A portion (3 mL) of 175 the diluted sample was combined with 1 mL of modified Wade reagent (0.03% 176 FeCl3.6H2O + 0.3% sulfosalicyclic acid) in a 15-mL falcon tube, thoroughly mixed 177 on a vortex and absorbance of colour reaction determined at 500 nm using a UV 178 spectrophotometer (UVLine 9400, Schott Instruments, France). 179 A standard curve was prepared using series of calibration standards containing 0, 180 1.12, 2.24, 3.36, 5.6, 7.84, or 11.2 mg L−1 PA-P from phytic acid sodium salt 181 hydrate (Sigma, P8810) and phytic acid content determined as above. 182 2.5 183 Means and standard deviations for experimental data were computed using Statistica 184 7.0 (Tulsa, OK, USA). Response surface methodology (RSM) was used to relate 185 product characteristics to extrusion variables. Response surface plots were generated 186 as a function of two variables, while keeping the third variable constant at its 187 intermediate value. Regression coefficients were generated by a second order 188 polynomial: 189 𝑌𝑖 = 𝐵𝑜 + 𝐵𝑖 ∑3𝑖=1 𝑋𝑖 + 𝐵𝑖𝑖 ∑3𝑖=1 𝑋𝑖 2 + 𝐵𝑖𝑗 ∑ ∑3𝑖,𝑗=1 𝑋𝑖 𝑋𝑗 + 𝜀 190 191 Statistical analysis and process optimization (2) where, 𝑌𝑖 is a response variable;𝐵0 is a constant; 𝐵𝑖 , 𝐵𝑖𝑗 are gradients; 𝑋𝑖,𝑗 are factors; and 𝜀 is error term. 192 The simultaneous optimization of the process conditions and product responses 193 (protein digestibility, WAI, WSI, extrudate expansion ratio (ER), bulk density and 194 hardness) for Roba1 bean snack was accomplished using a multi-response 8 195 desirability method (Derringer & Suich, 1980). Individual desirability functions for 196 each response variable were manipulated to achieve optimum values (Granato, 197 Ribeiro, Castro, & Masson, 2010). In this study, WAI, bulk density and extrudate 198 hardness demanded minimization, while protein digestibility, WSI and ER were 199 maximized. 200 201 9 202 203 RESULTS AND DISCUSSION 3.1 Extrudate physicochemical properties 204 205 Roba1 bean extrudate properties differed significantly with variation in die 206 temperature, feed moisture and screw speed (Tables 2, 3 and 4). The coefficients of 207 determinations (R2) for regression equations varied between 0.484 and 0.842 with 208 significant probability values (p < 0.05, p < 0.01 and p < 0.001) (Table 4). 209 Significant negative quadratic effects (p < 0.01) of die temperature and positive 210 interaction effects (p < 0.05) of die temperature with feed moisture on protein 211 digestibility were observed (Table 4). Total polyphenols were significantly 212 decreased with increase in feed moisture (linear terms, p < 0.05) and interactions of 213 die temperature (linear) with feed moisture (quadratic). Increases in screw speed 214 (linear terms) and feed moisture (quadratic terms) resulted in increases in total 215 polyphenols. The increases in screw speed (quadratic) and interaction effects of die 216 temperature (quadratic) with screw speed (linear) resulted in decrease in total 217 polyphenols. Possibly high screw speed limits the time of exposure of polyphenols 218 to heat destruction (Anton et al, 2008). 219 Phytates are considered undesirable in foods since they form complexes with major 220 divalent and trivalent cations (Ca, Fe, Zn, Mg and Cu) bearing an effect on mineral 221 uptake, and also bind with proteins affecting their nutritional quality and absorption 222 (Greiner, R., Konietzny, U., Jany, 2006). In this study, reduction in phytate content 223 (Table 3) by at least 12 % as a result of extrusion was observed. Raw Roba1 flour 224 was initially analyzed with average phytate content of 38.3 mg/g. No significant 225 effect of change in feed moisture, die temperature and/or screw speed (linear, 226 quadratic or interaction terms) on phytates were observed (p > 0.05) for Roba1 227 extrudate as seen from regression models. 10 228 Table 2 229 Values represent means of three replicates  standard deviations. Exp. run 1 2 3 4 5 6 7 8 9 19 11 12 13 230 11 Feed moisture (%) 15.0 17.5 17.5 20.0 15.0 15.0 17.5 20.0 20.0 15.0 17.5 17.5 20.0 Effect of extrusion conditions on Roba1 extrudate nutritional properties Die Temp. (°C) 120 120 120 120 135 135 135 135 135 150 150 150 150 Screw speed (Hz) 40 35 45 40 35 45 40 35 45 40 35 45 40 Protein content (g/100g) 23.490.01 24.040.34 24.040.07 24.880.78 22.261.37 24.290.41 24.050.05 24.000.13 24.100.04 24.160.49 23.590.08 24.720.40 24.220.31 Protein digestibility (%) 82.330.48 82.890.52 82.160.90 80.591.80 83.400.10 81.440.55 83.640.43 82.640.76 82.340.77 82.411.26 81.080.30 81.590.71 82.400.14 Polyphenol content (mg/100g) 27.863.12 23.724.92 24.151.65 23.150.53 22.720.87 31.221.11 24.344.20 17.912.66 20.201.64 30.730.55 24.534.42 18.582.12 35.711.73 Phytate content (mg/g) 28.891.39 28.521.34 30.411.70 25.892.65 27.992.11 31.023.48 30.491.83 29.150.41 32.123.46 26.384.47 33.583.24 29.480.16 33.011.87 231 Table 3 Exp. Feed run moisture (%) 15.0 1 17.5 2 17.5 3 20.0 4 15.0 5 15.0 6 17.5 7 20.0 8 20.0 9 15.0 19 17.5 11 17.5 12 20.0 13 232 233 234 Mean values of extrudate physical and functional properties Die Temp. (°C) 120 120 120 120 135 135 135 135 135 150 150 150 150 Screw speed (rpm) 40 35 45 40 35 45 40 35 45 40 35 45 40 WAI (g / WSI g) (mL/g) Radial expansion ratio (ER) 2.520.07 2.450.07 2.500.01 2.400.03 2.520.05 2.590.07 2.450.06 2.360.04 2.470.05 2.540.04 2.420.01 2.480.03 2.530.01 3.980.61 0.390.06 4.670.10 0.330.02 4.350.35 0.360.02 4.200.18 0.350.05 3.690.29 0.410.03 3.470.08 0.460.00 3.980.30 0.380.02 4.430.17 0.360.02 3.860.49 0.380.01 3.630.27 0.360.10 4.470.13 0.360.01 3.970.08 0.420.01 3.710.30 0.420.03 Values represent means of three replicates  standard deviations 1 Final pasting viscosity determined at 95°C using RVA * Area F-D, Area of force deformation curve, (cm)2 12 Bulk density (g/cm3) 1.500.06 2.060.05 1.810.08 2.170.33 1.490.25 1.330.05 1.840.04 1.850.58 1.780.01 1.580.31 2.150.05 1.630.04 1.720.07 Peak force (N) [Area F-D (cm)2 * 30.221.48 46.175.71 34.081.63 48.201.28 32.661.84 25.621.15 35.653.67 35.174.85 29.503.43 26.384.79 43.103.33 29.780.58 31.862.86 123.724.93 158.5810.62 132.645.79 172.828.74 130.855.35 110.143.72 140.319.36 137.3513.76 120.668.90 112.9013.79 159.038.39 122.172.70 131.4020.09 Mean force Final (N) viscosity1 (cP) 16.500.68 259.0022.00 21.820.65 323.677.38 17.491.07 307.2016.80 23.291.21 333.5024.10 17.470.74 302.7010.90 14.710.50 298.8023.70 18.721.25 286.949.27 18.331.84 335.8014.70 16.341.31 292.009.02 15.081.84 298.0033.40 21.231.14 320.5045.90 16.310.38 278.7028.40 17.542.70 308.8019.50 235 236 Table 4 Regression coefficients of the linear, quadratic and interaction effect of feed moisture die temperature and screw speed on Roba1 extrudate properties 237 Variable Constant Die temperature (X1) Feed moisture (X2) Screw speed (X3) X1*X1 X2*X2 X3*X3 X1*X2 X1*(X22) (X12)*X2 X1 X3 (X12)*X3 X2*X3 R2 * 238 239 240 ** *** 241 13 Protein digestibility Total polyphenols 83.640*** -0.597 0.038 -0.566 -1.117** -0.589 -0.597 0.432 1.070* -0.474 0.308 0.513 0.415 0.484 Significant at p < 0.05 Significant at p < 0.01 Significant at p < 0.001 WAI WSI 24. 823*** -1.171 3.983*** -0.060 0.384*** 0.022* Radial expansion ratio -2.451*** -0.011 -3.623** 3.032** 1.979 2.561* -4.038*** 2.425* -5.028** 3.691* -1.578 -4.396* -1.219 0.653 0.283** -0.198* 0.147 -0.257** 0.136 -0.195* -0.142 -0.048 0.004 -0.072 -0.085 0.651 -0.031** 0.003 -0.019 0.014 0.003 0.023* -0.013 0.035* 0.005 0.003 -0.001 0.499 -0.069*** 0.042** -0.001 0.035* 0.013 0.029 0.049* 0.009 0.004 0.012 -0.013 0.645 Bulk density 1.837*** -0.024 Hardness (Peak Force) 35.647*** -1.844 0.204** -0.120 0.054 -0.197** -0.029 -0.131* -0.070 -0.001 -0.069 -0.138 0.024 0.659 1.600 -3.179* 3.033* -4.512** -0.398 -3.125** -3.201* 4.264** -0.306 -3.172* 0.342 0.842 Area F-D Mean force 140.310*** -2.505 18.723*** 0.442 4.256 -9.348** 6.628* -11.728*** -3.834 -7.651* -10.552* 12.644** -2.731 -6.352 1.005 0.803 -0.623 -1.186* 0.381 -0.939* -1.559*** -1.083** -1.350* 1.689** -0.147 -1.127* 0.190 0.817 242 Extrudate expansion and bulk density 243 Increases in feed moisture by linear terms resulted in reduced expansion of Roba1 244 extrudate, while increases in screw speed (linear), feed moisture (quadratic) and 245 interaction terms of die temperature with feed moisture resulted in increased 246 extrudate expansion (Table 4). Decreases in expansion at high feed moisture might 247 be due to reduced elasticity of dough through plasticization of melt in the extruder 248 (Ding, Ainsworth, Plunkett, Tucker, & Marson, 2006; Hagenimana, Ding, & Fang, 249 2006). The significant increases in expansion (p < 0.01) due to increased screw 250 speed may be attributed to bubble growth resulting from increased water vapour 251 pressure at the die nozzle (Hagenimana et al., 2006). The results suggest that low 252 feed moisture and high screw speed would be important for attainment of high 253 expansion of extrudates. 254 Bulk density is related to the extent of extrudate expansion (Hagenimana et al., 2006) 255 and is very important in the production of expanded and formed food products. 256 Increases in feed moisture at low die temperatures (<130 °C) and high feed moisture 257 at high die temperatures (>145°C) resulted in increased bulk density (Table 3; Fig. 258 1). High bulk density at high feed moisture was probably due to the lubricating and 259 plasticizing effect of water, which lowers the mechanical shear effects and disruption 260 of starch in the extruder (Altan & Maskan, 2011). High bulk density could also be 261 due to rupture of the starch cell walls caused by fibre particles before the gas bubbles 262 in the starch attained full expansion (Altan & Maskan, 2011; Chiu, Peng, Tsai, Tsay, 263 & Lui, 2013). Chiu et al (2013) reported that high bulk density may result from 264 binding of water to non-starch polysaccharides (fibre) which inhibits water loss at 265 the die thus reducing expansion. Low bulk density at high screw speed (Fig. 1b and 266 1c) for all temperatures was probably due to starch gelatinization and increased 14 267 expansion of extrudate caused by increased water vapour pressure at the die 268 (Hagenimana et al., 2006). 269 270 271 272 Figure 1a to 1h Response surfaces curves of bulk density, WAI and WSI as affected by die temperature, feed moisture and screw speed 273 In this study it is notable that high bulk density was a function of increase in feed 274 moisture (linear terms) and the interactions of feed moisture (linear terms) with die 15 275 temperature (Table 4). Low feed moisture and high screw speed therefore would be 276 necessary for obtaining an extrudate with low bulk density. 277 Water absorption and solubility index 278 Water absorption index (WAI) was significantly influenced by feed moisture in both 279 linear and quadratic terms, screw speed (linear terms) and the interaction of die 280 temperature and feed moisture (Table 4, Fig. 1d to 1f). The negative coefficients of 281 the linear terms of screw speed, quadratic terms of feed moisture and interaction 282 terms of die temperature and feed moisture indicate that WAI decreases with 283 increase of those variables. Positive coefficients of the linear terms of feed moisture 284 indicated that WAI increased with increases in feed moisture. The increase in WAI 285 with increasing feed moisture could be attributed to the dispersion of starch in excess 286 water, the increased degree of starch damage by gelatinization and the extrusion 287 induced fragmentation of starch granules (Chiu et al., 2013; Ding, Ainsworth, 288 Tucker, & Marson, 2005; Hagenimana et al., 2006; Yagcı & Gögüs, 2011). High 289 moisture content builds low viscosity, thus allowing for internal mixing and uniform 290 heating of the dough which would account for enhanced gelatinization (Yagcı & 291 Gögüs, 2011). Additionally, protein denaturation, starch gelatinization and swelling 292 of fibre at high feed moisture could be responsible for increase in WAI (Altan & 293 Maskan, 2011). 294 Increases in die temperature (linear) and interactions of die temperature and feed 295 moisture (linear and quadratic terms) caused increase in WSI, while increases in feed 296 moisture (linear) resulted in reduction of WSI (Table 3). Similarly increases in 297 screw speed caused significant increase in WSI at low feed moisture and at high die 298 temperature (Fig. 1h and 1i). Increase in WSI at high die temperatures was probably 299 associated with the disintegration of starch granules and low molecular compounds 300 from extrudate melt during the extrusion process, thus increasing the soluble 16 301 material.(Yağcı & Göğüş, 2008) The increase in WSI with increase in screw speed 302 was in agreement with previous works (Altan & Maskan, 2011; Altan et al., 2008) 303 and may be due to increased specific mechanical energy, mechanical shear and 304 degradation of macromolecules. Increased WSI at high temperature and high screw 305 speed (Fig. 1h) was possibly associated with the starch degradation at high 306 temperature and greater shear action at high screw speed (Altan & Maskan, 2011; 307 Seth, Badwaik, & Ganapathy, 2015) 308 High WSI at high feed moisture could be explained by the complete changes of food 309 components from native forms, that is starch gelatinization and protein denaturation, 310 respectively (Yu, Liang; Ramaswamy, 2012). The low WSI at high feed moisture 311 and low die temperature, as well as at low screw speed and low die temperature 312 could be explained by the low degree of starch transformation, the reduced shear 313 degradation of starch and the low tendency to dextrinization (Hernandez-Diaz, 314 Quintero-Ramos, Barnard, & Balandran-Quintana, 2007; Liu et al., 2011). High 315 moisture content in extrusion processes may reduce protein denaturation and starch 316 degradation (Hernandez-Diaz et al., 2007). Alternatively, high feed moisture acts as 317 a plasticizer thus hindering full expansion and rupture of starch in the extruder (Liu 318 et al., 2011). WSI is an indication of the ease of solubilization and extent of water 319 absorption of the cooked product (Altan & Maskan, 2011). While WAI measures the 320 volume occupied by starch after swelling in excess water, thus its integrity in 321 aqueous dispersions ( Ding et al., 2005). Both WAI and WSI can be useful indicators 322 of the suitability for use of extruded starchy products in suspensions or solutions 323 (Yagcı, & Gögüs, 2011). Results from this study were in agreement with Ding et 324 al., (2006) who reported increases in WSI as extrusion temperature increased at feed 325 moisture of 18.2%. High WAI obtained at moderate feed moisture and low-to-high 326 die temperatures possibly reflected the ability of Roba1 extrudate to absorb moisture 327 and the stability of starch polymer composites upon exposure to water treatment; 17 328 which would be a good attribute in flour applications for aqueous dispersions 329 (Sarifudin & Assiry, 2014) such as soups and gruels. This study therefore reveals 330 that both extrusion conditions and material composition appear to influence the 331 functional properties of bean extrudates (WAI, WSI, bulk density and expansion). 332 Extrudate texture 333 The hardness, crunchiness and crispiness of Roba1 extrudates measured in terms of 334 peak force and area under force deformation curve were determined. Peak force 335 represents the resistance of extrudate to initial penetration and is believed to be the 336 hardness of the extrudate (Anton & Luciano, 2007; Ding et al., 2006). The regression 337 model of extrudate hardness was significant as a function of screw speed (linear 338 terms), feed moisture (quadratic terms), and the interactions of feed moisture with 339 die temperature (Table 4). High screw speed and low feed moisture resulted in soft 340 extrudates. Increase in feed moisture at low die temperatures resulted in increased 341 extrudate hardness (Fig. 2 and 3). 342 343 344 Figure 4 Response surfaces showing effect of die temperature, feed moisture and screw speed on peak force (hardness) of Roba1 extrudate 345 18 346 347 348 Figure 5 - 6 Pareto charts showing significance of effects of extrusion conditions on Roba1 extrudate hardness and crispiness. 349 Positive coefficients of feed moisture (LT and QT) and interactions of die 350 temperature (QT) with screw speed resulted in increased extrudate hardness (Table 351 4, Fig. 3). This could partly be due to the reduced dough friction in the extruder, 352 permitting rapid extrusion and full expansion of extrudate at the die exit at high 353 temperatures (Hagenimana et al., 2006). Increase in screw speed (LT) significantly 354 (p < 0.05) decreased Roba1 extrudate hardness, particularly at low feed moisture 355 and high die temperature (Figures 2b, 2c, and 3). Low hardness, a favoured property 356 of extrudates (Meng et al., 2010), was observed at low feed moisture and high screw 357 speed. Results from this study are similar to reports from previous works (Altan et 358 al., 2008; Ding et al., 2005), which showed that extrudate hardness increased with 359 increase in feed moisture at low die temperatures. This may be due to the reduced 360 pressure resulting from lubrication of extruder walls at high feed moisture, thus 361 lowering expansion. The hardness of extrudates was highly related to the bulk 362 density (Fig. 1 and 2). Low screw speed resulted in harder extrudates, which 363 probably was due to increase in melt viscosity of the mix (Ding et al., 2006; Maskus 364 & Arntfield, 2015). 19 365 The area under the force determination curve (Table 2) represents the energy 366 required for a given displacement and can measure the crispiness and brittleness in 367 texture of a product (Anton & Luciano, 2007). Roba1 extrudate brittleness tended to 368 increase with increase in screw speed (LT, p < 0.01) and decrease in feed moisture 369 (QT, p < 0.001)(Table 4; Figure 3b). Increase in die temperature (QT) increased the 370 crispiness of Roba1 extrudate. Interactions between die temperature (quadratic) and 371 feed moisture (linear) significantly (p < 0.05) decreased the area under the force- 372 deformation curve (Table 4; Fig 3b), thus increased the brittleness/crispiness of 373 extrudate. Die temperature (QT) and screw speed (LT) interaction significantly (p < 374 0.05) increased the extrudate crispiness. The results suggested that low feed moisture 375 (QT), high screw speed (LT) and die temperature-feed moisture interactions during 376 bean extrusion result into soft crispy extrudates, which is in agreement with Altan 377 et al., (2008) and Ding et al., (2006) who reported similar findings. 378 379 3.2 380 Multi-response desirability optimization of Roba1 extrusion conditions and extrudate properties 381 Optimal extrusion conditions that simultaneously satisfied all selected extrudate 382 responses and could improve the efficiency of the process for Roba1 snack extrudate 383 were identified (Fig. 4). Notable, individual optimal desirability values (𝑑) (Fig. 4) 384 differed from those obtained with a global optimal desirability (𝐷). This was due to 385 the conversions that occurred statistically to reach a simultaneously optimal values 386 (Tumwesigye et al., 2016; Vera Candioti et al., 2014). 20 387 388 389 Figure 7 Desirability index and predicted response variables in multi-response optimization of Roba1 extrusion conditions and product properties 390 Results revealed that die temperature 142.5 °C, feed moisture 15 % and screw speed 391 45 Hz, were sufficient for producing best characteristics for Roba1extrudate with 392 bulk density 1.32 g/cm3, expansion ratio 2.60 and hardness 24.4 N at optimal 393 desirability of 0.75. High expansion, low bulk density and hardness were identified 394 as important attributes in extruded beans snack processing, properties also 395 established by previous workers (Altan & Maskan, 2011; Jyothi, Sheriff, & Sajeev, 396 2009; Maskus & Arntfield, 2015). These findings show that low feed moisture and 21 397 high screw speed are essential for obtaining bean snack extrudate with desirable 398 expansion, hardness, bulk density and protein digestibility. 399 400 CONCLUSION 401 The results of this study revealed that die temperature, feed moisture and screw 402 significantly influenced Roba1 bean extrudate properties. Regression coefficient 403 (R2) values showed significant effect of individual factors and their interactions in 404 influencing bean extrudate properties. Increases in WAI, bulk density and extrudate 405 hardness and reductions in expansion and WSI were a function of the linear, 406 quadratic and interaction relationships of feed moisture, die temperature and screw 407 speed. Low feed moisture and high screw speed were found essential to produce a 408 soft crispy snack extrudate with low bulk density and high water solubility index. 409 Die temperature 142.5 °C, feed moisture (15%) and screw speed 270 rpm were found 410 as optimal process conditions for producing an expanded snack extrudate from 411 Roba1 bean with optimal desirability >0.7. Results of this study therefore can be 412 very important for application in the food industry, providing extrusion conditions 413 that for generating acceptable and nutritious products. This in turn will contribute to 414 increased diversity of bean products on the market, reduce monotony and overcome 415 the challenge of hard-to-cook defects. Subsequent inclusion of Roba1 beans in 416 extruded snacks shall increase protein, fibre and mineral intake by the consumers. 417 418 419 22 420 Acknowledgement 421 The authors acknowledge the special funding by the African Development Bank 422 (ADB) to Government of Uganda through the Centre for International Tropical 423 Agriculture (CIAT). We gratefully acknowledge Peak Value Industries Ltd (U) for 424 providing the extrusion facility. 425 426 Highlights  Optimal extrusion conditions for Roba1 bean snack extrudate were 427 428 determined  Extrusion conditions linear, quadratic and interaction terms were 429 430 significant  431 432 High extrudate bulk density and hardness were associated with high feed moisture 433  High screw speed and die temperature resulted in increased extrudate 434 435 expansion 436 REFERENCES 437 438 Altan, A., & Maskan, M. (2011). Development of Extruded Foods by Utilizing Food 439 Industry By-Products. In Advances in Food Extrusion Technology (pp. 121– 440 168). CRC Press. https://doi.org/doi:10.1201/b11286-710.1201/b11286-7 441 Altan, A., McCarthy, K. L., & Maskan, M. (2008). Evaluation of snack foods from 442 barley–tomato pomace blends by extrusion processing. Journal of Food 443 Engineering, 84(2), 231–242. https://doi.org/10.1016/j.jfoodeng.2007.05.014 444 Anderson, J. A., Gipmans, M., Hurst, S., Layton, R., Nehra, N., Pickett, J., … 445 Tripathi, L. (2016). Emerging Agricultural Biotechnologies for Sustainable 446 Agriculture and Food Security. Journal of Agricultural and Food Chemistry, 23 447 64(2), 383–393. https://doi.org/10.1021/acs.jafc.5b04543 448 Anton, A. A., Fulcher, R. G., & Arntfield, S. D. (2009). Physical and nutritional 449 impact of fortification of corn starch-based extruded snacks with common bean 450 (Phaseolus vulgaris L.) flour: Effects of bean addition and extrusion cooking. 451 Food Chemistry, 113(4), 989–996. 452 Anton, A. A., & Luciano, F. B. (2007). INSTRUMENTAL TEXTURE 453 EVALUATION OF EXTRUDED SNACK FOODS: A REVIEW 454 EVALUACIÓN INSTRUMENTAL DE TEXTURA EN ALIMENTOS 455 EXTRUIDOS: UNA REVISIÓN. Ciencia Y Tecnologia Alimentaria, 5(4), 245– 456 251. https://doi.org/10.1080/11358120709487697 457 Anton, A. A., Ross, K. A., Beta, T., Fulcher, R. G., & Arntfield, S. D. (2008). Effect 458 of pre-dehulling treatments on some nutritional and physical properties of navy 459 and pinto beans (Phaseolus vulgaris L.). LWT-Food Science and Technology, 460 41(5), 771–778. 461 Berrios, J. D. J., Ascheri, J. L. R., & Losso, J. N. (2012). Extrusion Processing of 462 Dry Beans and Pulses. In Dry Beans and Pulses Production, Processing and 463 Nutrition 464 https://doi.org/10.1002/9781118448298.ch8 (pp. 185–203). Blackwell Publishing Ltd. 465 Bezerra, M. A., Santelli, R. E., Oliveira, E. P., Villar, L. S., & Escaleira, L. A. 466 (2008). Response surface methodology (RSM) as a tool for optimization in 467 analytical 468 https://doi.org/10.1016/j.talanta.2008.05.019 chemistry. Talanta, 76(5), 965–977. 469 Bouis, H. E., & Welch, R. M. (2010). Biofortification—a sustainable agricultural 470 strategy for reducing micronutrient malnutrition in the global south. Crop 471 Science, 50(Supplement_1), S-20-S-32. 472 Chiu, H.-W., Peng, J.-C., Tsai, S.-J., Tsay, J.-R., & Lui, W.-B. (2013). Process 473 Optimization by Response Surface Methodology and Characteristics 24 474 Investigation of Corn Extrudate Fortified with Yam (Dioscorea alata L.). Food 475 and Bioprocess Technology, 6(6), 1494–1504. https://doi.org/10.1007/s11947- 476 012-0894-6 477 Derringer, G; Suich, R. (1980). Simultaneous optimization of several response 478 variables. Journal of Quality Technology, 12(4), 214–219. 479 Ding, Q.-B., Ainsworth, P., Plunkett, A., Tucker, G., & Marson, H. (2006). The 480 effect of extrusion conditions on the functional and physical properties of wheat- 481 based expanded snacks. Journal of Food Engineering, 73(2), 142–148. 482 https://doi.org/10.1016/j.jfoodeng.2005.01.013 483 Ding, Q. B., Ainsworth, P., Tucker, G., & Marson, H. (2005). The effect of extrusion 484 conditions on the physicochemical properties and sensory characteristics of rice- 485 based expanded snacks. Journal of Food Engineering, 66(3), 283–289. 486 https://doi.org/10.1016/j.jfoodeng.2004.03.019 487 Gao, Y., Shang, C., Maroof, M. A. S., Biyashev, R. M., Grabau, E. A., Kwanyuen, 488 P., … Buss, G. R. (2007). A Modified Colorimetric Method for Phytic Acid 489 Analysis 490 https://doi.org/10.2135/cropsci2007.03.0122 in Soybean. Crop Science, 47(5), 1797. 491 Ghumman, A., Kaur, A., Singh, N., & Singh, B. (2016). Effect of feed moisture and 492 extrusion temperature on protein digestibility and extrusion behaviour of lentil 493 and horsegram. LWT-Food Science and Technology, 70, 349–357. 494 Granato, D., Ribeiro, J. C. B., Castro, I. A., & Masson, M. L. (2010). Sensory 495 evaluation and physicochemical optimisation of soy-based desserts using 496 response 497 https://doi.org/10.1016/j.foodchem.2010.01.014 498 methodology. Food Chemistry, 121(3), 899–906. Hagenimana, A., Ding, X., & Fang, T. (2006). Evaluation of rice flour modified by 499 500 surface extrusion cooking. Journal of Cereal Science, 43(1), 38–46. Hernandez-Diaz, J. R., Quintero-Ramos, A., Barnard, J., & Balandran-Quintana, R. 25 501 R. (2007). Functional properties of extrudates prepared with blends of wheat 502 flour/pinto bean meal with added wheat bran. Revista de Agaroquimica Y 503 Tecnologia de Alimentos, 13(4), 301–308. 504 HSU, H. W., VAVAK, D. L., SATTERLEE, L. D., & MILLER, G. A. (1977). A 505 MULTIENZYME TECHNIQUE 506 DIGESTIBILITY. Journal 507 https://doi.org/10.1111/j.1365-2621.1977.tb14476.x of FOR Food ESTIMATING Science, 42(5), PROTEIN 1269–1273. 508 Jain, Monika; Singh, Chetna; Gupta, Kushboo; Jain, P. (2014). Optimization of 509 Functional Food Ingredients and Their Processing Levels for Preparation of 510 Vermicelli Using RSM. International Journal of Engineering Sciences & 511 Research Technology, 3(6), 8–20. 512 Jyothi, A. N., Sheriff, J. T., & Sajeev, M. S. (2009). Physical and Functional 513 Properties of Arrowroot Starch Extrudates. Journal of Food Science, 74(2), 514 E97–E104. https://doi.org/10.1111/j.1750-3841.2008.01038.x 515 Korus, J., Gumul, D., & Czechowska, K. (2007). Effect of extrusion on the phenolic 516 composition and antioxidant activity of dry beans of Phaseolus vulgaris L. Food 517 Technology and Biotechnology, 45(2), 139. 518 Krupa-Kozak, U., & Soral-Śmietana, M. (2010). Bean seed proteins digestibility 519 affected by pressure and microwave cooking. Acta Alimentaria, 39(2), 234–238. 520 Liu, C., Zhang, Y., Liu, W., Wan, J., Wang, W., Wu, L., … Yin, Z. (2011). 521 Preparation, physicochemical and texture properties of texturized rice produce 522 by improved extrusion cooking technology. Journal of Cereal Science, 54(3), 523 473–480. 524 Makkar, H. P. S. (2000). A laboratory manual for the FAO/ IAEA co-ordinated 525 research project on “Use of Nuclear and Related Techniques to Develop Simple 526 Tannin Assays for Predicting and Improving the Safety and Efficiency of 527 Feeding Ruminants on Tanniniferous Tree Foliage”. 26 528 Maskus, H., & Arntfield, S. (2015). Extrusion processing and evaluation of an 529 expanded, puffed pea snack product. Journal of Nutrition & Food Sciences, 530 5(4), 1. 531 Meng, X., Threinen, D., Hansen, M., & Driedger, D. (2010). Effects of extrusion 532 conditions on system parameters and physical properties of a chickpea flour- 533 based 534 https://doi.org/10.1016/j.foodres.2009.07.016 snack. Food Research International, 650–658. 43(2), 535 Natabirwa, H., Muyonga, J. H., Nakimbugwe, D., & Lungaho, M. (2017). Physico- 536 chemical properties and extrusion behaviour of selected common bean varieties. 537 Journal 538 https://doi.org/10.1002/jsfa.8618 of the Science of Food and Agriculture. 539 Natukunda, S., Muyonga, J. H., & Mukisa, I. M. (2016). Effect of tamarind ( 540 Tamarindus indica L.) seed on antioxidant activity, phytocompounds, 541 physicochemical characteristics, and sensory acceptability of enriched cookies 542 and 543 https://doi.org/10.1002/fsn3.311 mango juice. Food Science & Nutrition, 4(4), 494–507. 544 Nyombaire, G., Siddiq, M., & Dolan, K. D. (2011). Physico-chemical and sensory 545 quality of extruded light red kidney bean (Phaseolus vulgaris L.) porridge. LWT- 546 Food Science and Technology, 44(7), 1597–1602. 547 Pedrosa, M. M., Cuadrado, C., Burbano, C., Muzquiz, M., Cabellos, B., Olmedilla- 548 Alonso, B., & Asensio-Vegas, C. (2015). Effects of industrial canning on the 549 proximate composition, bioactive compounds contents and nutritional profile of 550 two Spanish common dry beans (Phaseolus vulgaris L.). Food Chemistry, 166, 551 68–75. https://doi.org/10.1016/j.foodchem.2014.05.158 552 Rathod, R. P., & Annapure, U. S. (2017). Physicochemical properties, protein and 553 starch digestibility of lentil based noodle prepared by using extrusion 554 processing. 27 LWT - Food Science and Technology, 80, 121–130. 555 https://doi.org/10.1016/j.lwt.2017.02.001 556 Rocha-Guzman, N. E., Gallegos-Infante, J. A., Gonzalez-Laredo, R. F., Bello-Perez, 557 A., Delgado-Licon, E., Ochoa-Martinez, A., & Prado-Ortiz, M. J. (2008). 558 Physical properties of extruded products from three Mexican common beans 559 (Phaseolus vulgaris L.) cultivars. Plant Foods for Human Nutrition, 63(3), 99– 560 104. 561 Seth, D., Badwaik, L. S., & Ganapathy, V. (2015). Effect of feed composition, 562 moisture content and extrusion temperature on extrudate characteristics of yam- 563 corn-rice based snack food. Journal of Food Science and Technology, 52(3), 564 1830–1838. https://doi.org/10.1007/s13197-013-1181-x 565 Siddiq, M., Kelkar, S., Harte, J. B., Dolan, K. D., & Nyombaire, G. (2013). 566 Functional properties of flour from low-temperature extruded navy and pinto 567 beans (Phaseolus vulgaris L.). LWT - Food Science and Technology, 50(1), 215– 568 219. https://doi.org/10.1016/j.lwt.2012.05.024 569 Steel, C. J., Gabriela, M., Leoro, V., Schmiele, M., Ferreira, R. E., & Chang, Y. K. 570 (2012). Thermoplastic Extrusion in Food Processing. In El-Sonbati A (Ed.), 571 Thermoplastic elastomers (pp. 265–290). https://doi.org/10.5772/36874 572 Tunick, M. H., Onwulata, C. I., Thomas, A. E., Phillips, J. G., Mukhopadhyay, S., 573 Sheen, S., … Cooke, P. H. (2013). Critical Evaluation of Crispy and Crunchy 574 Textures: A Review. International Journal of Food Properties, 16(5), 949–963. 575 https://doi.org/10.1080/10942912.2011.573116 576 Yagcı, S; Gögüs, F. (2011). 12 Quality Control Parameters of Extrudates. In 577 Advances in Food Extrusion Technology. 578 Yağcı, S., & Göğüş, F. (2008). Response surface methodology for evaluation of 579 physical and functional properties of extruded snack foods developed from 580 food-by-products. 581 https://doi.org/10.1016/j.jfoodeng.2007.09.018 28 Journal of Food Engineering, 86(1), 122–132. 582 Yu, Liang; Ramaswamy, H. S. . B. J. (2012). Twin-screw Extrusion of Corn Flour 583 and Soy Protein Isolate (SPI) Blends: A Response Surface Analysis. Food 584 Bioprocess Technol, 5, 485–497. https://doi.org/10.1007/s11947-009-0294-8 585 29