Flour pasting properties define consumer selection decisions for boiled and flour-based cassava food products in Uganda

Enhanced adoption of new cassava varieties is dependent on the ability to breed for acceptability attributes that define boiled roots and flour-based meals. This study assessed the role of pasting properties in defining farmers’ acceptability. Cassava accessions in a Triadic Comparison of Technologies (TRICOT) approach were assessed for acceptability in addition to moisture content root softness (boiled roots)and pasting properties (flour). Results showed location based differences in moisture content (13-17%) and boiled root softness (0.5-6.0 N/cm). Pasting properties such as peak viscosity (4300-5500 cP), breakdown (2204-3234 cP) and final viscosity (2953-3493) also varied significantly among accessions. Linear relationships were observed between consumer acceptance parameters and pasting properties {(final viscosity and root mealiness (r=0.7), ease of mingling (r=0.6), product stickiness (r=0.5) and overall performance of the flour based meal (r=0.6)}.Therefore pasting property analysis can differentiate cassava accessions and is an important tool in selection of consumer acceptable varieties.


Introduction
Cassava's (Manihot esculenta Crantz) ever-increasing role in food and industrial value chains justifies interventions aimed at sustaining and/or increasing its production and productivity [1]. It's partly for this reason that significant efforts and resources have for the past decades been devoted to genetically improve cassava and enable it overcome acute production stresses and/or obstacles to optimal production [2]. These breeding efforts led to increased adoption of new varieties i.e., adoption increasing from 20% in 1993 to 80% in 1999 and is still increasing up to now [3].
However, some improved varieties like NASE1, NASE10, NASE 11 have since been abandoned with farmers choosing to revert back onto their local varieties which are known to have good root quality traits despite having poor agronomic performance [4] [5]. Consequently, this has been a major a major challenge to cassava breeding's relevance and thus necessitating a redesign of the breeding process and philosophy. Worse still is the hazy description of preferred end-user traits which makes it difficult to measure and henceforth to breed for [6]. In some instances, a specific end-user need is tagged to a range of traits which if not well understood cannot be bred and/or selected [3]. On the other hand, some traits related to the processed product and/or the processing conditions, are definitive of the acceptability of a particular product from a variety [7]. Such traits may not be readily measured in the raw primary product.
Traits such as the cooking time, boiled root quality, compositional profiles of processed flour, the cooking properties and their interactions, drive acceptability [8] [9][10] [11]. Leveraging on lessons elsewhere, preferred and non-preferred quality traits in raw, processed and final-end product have been compiled in Uganda to inform breeding decisions [12]. However, before prioritizing such traits, it is important to know how these traits are measured, and understand the relationships between such traits and sensorial properties upon which selection decisions are made. With some traits being associated with gender and/or group-based livelihood strategies [12], the inclusion of such traits in variety development is increasingly becoming critical. Triadic comparisons of technologies (TRICOT) provides for a farmer-led pro-active approach that enables inclusive and unbiased participation of farmers in variety selection decisions [7]. TRICOT allows the interrogation of farmer preferences and enables each farmer to identify the 'best' and 'worst' variety in aspects of agronomy, processing, culinary traits and overall performance conditions. Thus in this study, we adopted the TRICOT approach in evaluating the food products arising from our elite cassava accessions. Specifically, the objective of this study was to determine the pasting properties of flour obtained from farmers participating in TRICOT trials alongside farmer led acceptability assessments of the accessions used. This was meant to provide a definitive understanding of pasting properties as measurable traits that define consumer preferences for boiled cassava roots and flour-based meal in Uganda. The influence of these traits on acceptability of a particular variety among farmers was also assessed.

Study area description
This study targeted regions were "boiled roots" and/or 'flour-based meal" are primary cassava food products consumed by residents [7]. Accordingly, for 'flour-based meal" the districts of Serere and Kaberamaido (eastern region), and Arua (West Nile region) were selected. For "boiled roots" four districts namely Luwero and Mityana (central region), and Serere and Kaberamaido (Eastern region), were selected. Thus, Serere and Kaberamaido served as comparative locations for both cassava products under assessment. In each of these districts, farmer managed variety selection trials were established following the TRICOT procedure where randomization is maximized among farmers. Briefly, two counties leading in cassava production were selected per district with help of district agricultural officers. From each county, four parishes were selected while 10 farmers were selected per parish. Each farmer was randomly assigned three accessions coded as "A" "B" and "C". These were established on small plots that comprised 18 plants planted in three-meter rows of six plants per row. Farmers were encouraged to maintain their plots weed-free throughout the experimentation period that lasted 12 months when harvesting and processing of food products was done.

Consumer's acceptability tests on "boiled roots"
At each farmers field (Mityana, Luwero Arua and Kaberamaido and Serere) five plants were harvested per accession. All roots harvested from each plot were separately placed in plastic bags labelled "Plot A" "Plot B" and "Plot C" and carried to the farmer's home. Three appealing roots were then selected per accession, peeled and washed. Tapering ends of the root (proximal and distal) were chopped off and the remaining section peeled, washed in clean water and sectioned into small pieces according to routine farmer's practices.
Thereafter, the chopped pieces were separately placed in three saucepans labelled "Plot A" "Plot B" and "Plot C". The pans were covered with banana leaves and cooking was then done concurrently over firewood. To allow for fair comparison of accessions, three fire places were lit concurrently as done routinely by farmers. Cooking was done for 45 minutes. After 45 minutes, the boiled roots were served and assessed for sensory and softness attributes.
Accordingly, each farmer assessed boiled root acceptability of the three accessions harvested at his/her site for cooking time, mealiness, taste, softness, fibrousness and overall liking [12] [7]. This assessment involved defining the "best" and "worst" accessions per attribute as outlined in TRICOT methodology. Furthermore, each accession was assessed for softness using penetrometer [12]. The probe of the penetrometer was pressed into each cooked piece to a depth of 1 cm and the force recorded in N/cm. Twelve (12) readings were recorded per accession.

Consumer's acceptability tests on "flour-based meal"
From the roots harvested in the districts of Kaberamaido, Serere and Arua, assessment of flour product was done alongside boiled cassava. At least 2-3 marketable roots were picked and approximately three-quarters of the total roots harvested per plot were peeled, washed, sliced into thin sections according to the farmer's routine practice. The slices were dried for three days. The dried chips were then milled into flour per plot labelled as "A", "B" or "C". The acquired flour was then prepared into cassava flour based meal. Briefly, the farmer boiled enough water to make three 'flour-based meals" tagged to each of the plots A, B and C. Once the water boiled, the household member entrusted with preparing meals took turns to make a meal from each accession, ensuring that similar quantity (1 kg) of flour and water was used for each accession. The prepared "flour-based meal" was then served on three separate pre-labeled plates.
Acceptability assessments of "flour" by farmers were based on ease of flour drying and ease of mingling. The "flour-based meal" was also assessed for color, texture and taste. Just like for boiled roots, individual farmer's ranked "best" and "worst" accessions. Data from different farmers was then aggregated on a 10 point scale as a description of each parameter and average scores of each variety across different farmers were generated.
Remnant flour samples tagged to respective plots ("Plot A" "Plot B" and "Plot C") from farmers were shipped to the laboratory for further analysis. Firstly, assessment was made for moisture content using a modified procedures [13]. Secondly, assessment for flour pasting properties was undertake using a rapid viscosity analyzer (RVA-4500, Perten Instruments, Australia) and the Thermocline for Windows Software, using a modified flour profile as described. For pasting properties, three grams of flour were weighed up into the RVA canister and a volume equivalent to 25 g of distilled water added. The RVA test profile was held constant at 50 °C, 960 rpm mixing speed for 10 seconds. The mixing speed was decreased to 160 rpm and the temperature held at 50 °C for an extra 50 seconds. The temperature was then steadily increased to 95 °C within 4 minutes and held constant at 95 °C for 2.5 min and steadily decreased back to 50 °C in 4 minutes. The final viscosity was documented after 13 minutes. This analysis enabled quantification of flour viscous load on pasting; flour ability to withstand heating and shear stress; time to attain peak viscosity; flour process-ability temperature; degree of re-association of pasted flour during cooling and tendency to formation of gel after cooking.

Data analysis
Data on acceptability of "boiled roots" was aggregated on the basis of accessions and the districts in which those accessions were grown. For each of the districts, a summary of performance of each accession was generated for each farmer. With each farmer growing three selected accessions, the best ranked accession for the farmer was assigned "score" three (3) while the worst ranked accession was assigned "score" one (1). Accessions ranked as medium in performance were thus assigned "score" two (2). The assigned scores were then used to select the best and worst performing accessions based on the ranking by farmers. Similarly, data on acceptability of "flour-based meal" was also aggregated as for the boiled roots according to the district of origin. Each farmer was allowed to identify the best and worst performing accessions based on the ranks. These ranks were later converted into "score" by assigning "score" three (3) to the best ranked, "score" two (2) to the medium performing accession and "score" one (1) to the worst ranked. Further, regression analyses were conducted to test relationships between: a) softness assessment using penetrometer and sensory attributes of 'boiled roots" and "flour-based meal"; and b) pasting properties generated from RVA and acceptability attributes. Correlations between pasting properties, accession performance attributes (scores) and sensory attributes (scores) were derived using the cor-function using R statistical software.

Results
The results obtained from the various analyses conducted are described. They point out the specific differences on comparison among the accessions used in this study. Results also point out the performance of the different accessions in relation to NAROCASS 1 the best variety of choice among farmers at the moment.

Cassava flour moisture content:
The moisture contents of flour samples prepared by farmers in the study areas ranged from 11-17%. There were no significant variations (p>0.05) observed in accessions of cassava used (Table 1). Accessions MM16/0707 registered the highest moisture contents (15.1%) while accession UG120024 registered the lowest moisture contents (13.75%) compared to other accessions and across all the study sites. When study sites were compared, it was observed that farmers in Arua had a lower average moisture content of 13.7% compared to farmers in Kaberamaido (15.02%) and Serere (14.53%). The moisture contents were higher than the recommended for cassava flour in all cases [14][9].

Acceptability rankings of "boiled roots" and "flour-based meals" cassava products
Results for acceptability of the flour based meal and boiled roots are presented in Table 3.
For the flour based meal, the overall best ranked accession was UG120193 followed by NAROCASS 1. The worst ranked accession over all was UG130016. Location specific differences were also observed for the flour meal based product with best ranked accession in Serere being UG120193 while it was UG120156, UG120193 and UG120198 in Kaberamaido. On the other hand, the best ranked accession for the flour based meal in Arua was UG120193 (Table 3) Over all, the best ranked varieties for boiled root product was UG120193 and UG120124 while the worst ranked varieties were MM16/1627, MM06/123, MM16/0707 and UG130016. In the individual districts, ranking was also different with the NAROCASS 1 as the best ranked variety in Serere, UG120156 as the best in Mityana and Kaberamaido while in Arua it was UG120193. These differences show differences in preference for the boiled root cassava product in specific areas (Teeken et al. 2018). Note: The value zero (0) denotes that the sample was not assessed in that particular instance.
Location wide comparisons for the performance of the boiled root showed that the most preferred accession over all was UG120193 followed by NAROCASS1 and UG120156. The two accessions (i.e., UG120193 and UG120156) can therefore be considered as replacement varieties for NAROCASS 1 the bench mark variety in this case. In the case of the flour based meal, accessions UG120193, UG130007 and UG120156 and NAROCASS 1 were the most acceptable. In this case too, the accessions UG120193, UG130007 and UG120156 can be considered to replace NAROCASS 1 (Figure 1).

Pasting properties of cassava flour
The definitive behavior of cassava flour during pasting was established as presented Table 4 and in Figure 2. Variability within specific accessions grown in different places and used in the study was low much as variety to variety comparisons showed significant differences. Flour peak viscosity ranged from 4300-5600cP representing a difference of 1200cP where the reference variety NAROCASS1 with a peak viscosity of 5100cP was in the mid-range and lower than for accessions UG120124 and UG120198. These variations underpin differences in flour components during cooking[8] [9].
Trough viscosity, an indicator of the ability of "flour-based meal" not to collapse during cooking [15], ranged from 2000-2720cP (Table 4). This indicated the differences in holding strength of flour during mingling of the "flour-based meal" for the accessions studied. Breakdown viscosity on the other hand ranged from 2200cP to 3250cP and thus being a highly variable quality parameter. In this study, accession UG120156 had the lowest breakdown viscosity while accession UG120198 was the highest, implying that accession UG120198 had the most stable paste. Accessions with high breakdown viscosity have critical usability and process-ability characteristics as has been obeserved in cassava [9] and other flour sources [15][16] [17] making them important for use where stable pastes are required. In this study, final viscosity ranged from 3000cP to 3500cP with clear differences between most of the accessions. The final viscosity shows the restructuring of starch molecules within the paste after cooling [15]. Thus different accessions used in this study restructed differently after cooling producing different characteristic pasting profiles. Like the final viscosity, set back viscosity measures the properties of the final product after cooling and is thus highly related to acceptability criteria of the flour based meal and the boiled root based meals. In this study, the setback viscosity ranged from 820cP to 1120cP in the different accessions, showing significant variations (p<0.05) among accesions. Pasting temperature is associated with onset of starch gelatinization and thus describes temperature for maximum absorption of water by the starch [11]. In this study, pasting temperatures ranged from 70.

Relationship between "flour-based meal" acceptability and pasting properties
Selection of the best "flour-based meal" from the evaluated accessions would ultimately depend on consumers and/or farmer's preferences [15] [3], which is very likely to be subjective and highly variable among households [3]. Thus in this study, we hypothesized that through quantifying pasting properties we could identify a metric that can corroborate consumers and/or farmer's preferences for "flour-based meal" product. Thus relationships between flour based meal acceptability parameters and the pasting properties were assessed using correlation analysis.
Peak viscosity had a positive significant relationship with "flour-based meal" stickiness (r =0.5, p = 0.05), and with overall performance (r=0.5, p =0.05) showing that "flour-based meal" stickiness can be quantified using peak viscosity as a metric. As such, high peak viscosity was observed in accessions UG120124, UG120193, UG120198 which also had high stickiness rankings. However, weak negative correlations (r =-0.2, p= 0.1) were observed between peak viscosity and "ease of mingling" as it would be hard to mingle a very thick paste (Table 6). On the other hand, the trough viscosity was significantly correlated to taste (r=0.5, p =0.05), performance of the flour based meal (r=0.6, p =0.05) and the choice of the best accession as ranked by consumers (r=0.5, p =0.05) ( Table 5).
Final viscosity had a positive correlation with "flour-based meal" texture (r=0.5, p =0.05), taste (r=0.5, p =0.05) and overall performance of the flour (r=0.7, p =0.05). In addition, the final viscosity was significantly correlated to the overall performance (r=0.5, p =0.05). On the other hand, Set back viscosity was positively related to "ease of mingling" (r=0.5, p =0.05) confirming the observation that increased viscosity makes food processing harder to accomplish (Table 5). Overall, the best performing accession for "flour-based meal" was UG120193, while the worst performing accession was UG130016. In terms of pre-consumption attributes, UG120124 was the best owing to its rapid drying, while UG120198 was good for mingling.

Relationship between "boiled roots" acceptability and pasting properties
Like for the flour acceptability parameters, boiled root acceptability parameters were also correlated to pasting prorties. Peak viscosity was linearly related to mealiness (r =0.7, p =0.05), cooking time (r=0.5, p =0.05) and sensorial softness (r=0.5, p=0.05). Cassava accessions UG120124, UG120193, and UG120198, characterized with higher peak viscosity were considered mealy by consumers. Trough viscosity on the other hand, was positively correlated with cooking time (r=0.4, p =0.05), mealiness (r=0.5, p =0.05), softness (r=0.6, p =0.05) and the overall performance of boiled roots (r=0.6, p=0.05). Like peak viscosity, breakdown viscosity was positively related to cooking time (r=0.5, p=0.05), mealiness (r=0.7, p=0.05), softness as consumer perception (r=0.5, p =0.05) and fibrousness (r=0.4, p =0.05).  4.0. Discussion: 6 Flour and boiled root acceptability in TRICOT based studies were studied. The TRICOT 7 methodology allowed for a community led intervention, where farmers could also be 8 engaged in the proper management procedures for producing standard products from 9 cassava. Based on the results, differences in preferences were observed across study 10 locations. The observed differences in study locations have been observed in related 11 studies [6] and indicate the need for location specific breeding targets [3]. It also indicated 12 that breeding programs need to provide accompanying packages detailing the 13 processing and utilization of different cassava varieties. 14 The observed location based differences in moisture content pointed to differences in 15 flour processing methods among different cassava growing communities [19]. Besides 16 environmental differences in such locations may influence moisture contents given the 17 hygroscopic nature of cassava flour [14]. This further underlines the need for location 18 specific processing strategies to deliver standard cassava products. Location based 19 differences were also observed for root softness. The observation could be attributed to 20 environmental effects on cassava root properties with specific differences in soil factors 21 influencing root properties [20]. 22 The variability observed for root softness in most varieties however pointed to the 23 possibility of using root softness as reliable metric to undertake selection in cassava [2]. 24 However, this should be coupled to explainable variations between varieties as was the 25 case in this study. More importantly, clones UG120198 and UG120024 which had low 26 softness scores could be selected since farmers in Uganda are interested in varieties that 27 have softer roots [7]. 28 Location based differences in acceptability of the cassava product were observed in this 29 study. The differences were expected given the different socioeconomic contexts in 30 utilisation of cassava as a crop [21]. But most importantly, the study was able to identify 31 accessions UG120193 and UG120156 as replacement varieties for the bench mark variety 32 NAROCASS 1. This was based on the higher or equal to NAROCASS 1 performance 33 displayed by these accessions in terms of their boiled root and flour meal properties. 34 Such performance therefore warrants further advancement for possible release. 35 The significant differences (p<0.05) observed for final viscosity among the accesions 36 shows that final viscosity can be used as a metric to define extent of starch 37 retro-gradation. This is based on the differences in amylose molecular re-association 38 [15] during cooling which leads to formation of viscous pastes on cooling. The differences 39 therefore are related to the intercation between amylose content and other flour 40 compistes and are measured as the setback viscosity [15] [16]. Some accessions with higher 41 peak time were observed such as UG120198 which indicate extended times for forming 42 pastes and hence longer processing times [22]. The peak may thus influence the cooking 43 time of boiled roots or the time taken to produce the cassava based meal much as it may 44 not be the direct measure of cooking time [16]. 45 The measurement of viscosity parameters provided for a clear understanding of some of 46 the consumer based preferences of both the flour and boiled root based meals. Peak 47 viscosity was found to be indicative of the stickness of the bread where highly viscous 48 pastes were also sticky in nature. This was indicated by the positive significant 49 relationships (r=0.5, p =0.05) observed between peak viscosity and flour based meal 50 stickiness. The positive significant relationships (r=0.6, p =0.05) observed between 51 performance of the flour based meal and trugh viscosity indicated that performance of 52 the flour based meal was perceived as a processing property [1][7] by the consumers and 53 thus could be easily measured from the ability of the paste not to "collapse" during 54 preparation/cooking. The high positive and significant relationship observed between 55 final viscosity and flour meal over all performace is due to the fact that the "flour-based 56 meal" is consumed on cooling [7]. Therefore, final consumer perception is very likely to 57 be influenced by the final viscosity. 58 59 For the boiled root based meal, peak viscosity was significantly related to boiled root 62 mealiness. Mealiness is an essential attribute preferred for boiled cassava roots[7] [3]. 63 Ironically, it's difficult to quantify and henceforth breed for [7]. Findings presented herein 64 suggest that high peak viscosity can within limits be used to infer a better degree of 65 mealiness. 66 However, some pasting parameters were not definitive of any consumer preference 67 based characteristics. For example, there was no relationship observed between 68 breakdown viscosity and consumer acceptability attributes. It suffices to note that break 69 down viscosity defines paste stability during the mingling process [9], a phenomenon that 70 may not be adequately assessed from the final flour product, "flour-based meal". 71 Therefore, breakdown viscosity could not be used to inform selection decisions. In 72 addition, it calls for possible assesement of these traits by the consumer even through the 73 mingling process where possible. Since breakdown viscosity is related positively and 74 significantly related to peak viscosity (r=0.9) and other pasting properties (Table 6), 75 inferences for consumer based understanding can be based on peak viscosity as a metric. 76 The results obtained in this study further suggest that softness-related parameters which 77 also include mealiness, could be an indirect measure of stability of root components after 78 boiling [7]. This is further supported by the positive correlation between final viscosity 79 and overall performance of "boiled root" (r=0.6, p =0.05), since a more viscous and stable 80 matrix in the root would influence the observed softness score. Negative correlations 81 were observed between pasting temperature and cooking time (r=-0.5, p =0.05), mealiness 82 (r=-0.5, p =0.05) and overall performance of boiled root (r=-0.4, p =0.05). This shows that 83 higher pasting times result from increased time for processing the root a factor that is 84 inversely related to good root performance. Such relationships reveal that pasting 85 properties have important associations with cassava root quality attributes as perceived 86 by consumers. The findings also point to the possibility of using pasting properties for a 87 better description of the sensorial properties of cassava root and flour based meals. 88

89
From the acceptability analyses undertaken, accession UG120193 was the most 90 acceptable accession among the accessions tested. However comparative acceptability 91 was observed for NAROCASS 1, UG120156 and UG130007. These accessions had 92 comparatively high peak viscosity (~>5500 cP) and moderate peak times of 4.2 minutes 93 showing that they had similar pasting behavior. Thus acceptability can in part be 94 explained by the pasting behavior of flour from different varieties. Pasting properties 95 were linearly related to some boiled root properties such as mealiness, cooking time and 96 softness. These pasting properties can be used to understand root based manifestations of 97 acceptability after boiling and hence selection of varieties with acceptable boiled roots. 98 From the study, it was observed that according to farmers, the most acceptable accession 99 over all was UG120193 followed by UG120156. The two accessions performed equally to 100 or higher than NAROCASS 1 and can therefore be considered as replacement varieties 101 for the bench mark variety in this case. Pasting properties and viscosity parameters 102 related to the processing of cassava flour included the peak viscosity, trough viscosity, 103 and the final viscosity. These are relevant metrics for assessing the acceptability of the 104 boiled root and flour based meal. Thus they can also be used to define theacceptability of 105 particular varieties for flour based meal production.