URBAN CONSUMER ACCEPTABILITY OF TRADITIONAL (WET) AND DRIED NIGERIAN FUFU

URBAN CONSUMER ACCEPTABILITY OF TRADITIONAL (WET) AND DRIED NIGERIAN FUFU Tomlins, K. I., Sanni, L. O., Oyewole, O. B., Dipeolu, A. O., Ayinde, I. A., Adebayo, K., Wandschnieider, T. S.,, White, J. L., Westby, A. Natural Resources Institute, The University of Greenwich, Central Avenue, Chatham Maritime, Kent ME TB, United Kingdom INTRODUCTION University of Agriculture, P.M.B. 0, Abeokuta, Nigeria Cassava (Manihot esculenta) production in Nigeria is estimated to be million tonnes per annum and has dramatically increased since 986. Fufu is a fermented cassava product that is traditionally consumed in Nigeria and other West African countries. It is prepared by peeling roots, washing, cutting into pieces and steeping in water to ferment for - days. The fermented mash is pounded or simply washed over a fine sieve to remove fibre and the water expelled from the starch extract by pressing through muslin cloth to produce fufu (Oyewole and Sanni 995). Fufu is considered by consumers to be of good quality when it has a smooth texture, characteristics aroma and is creamy-white, grey or yellow in colour (Akingbala et al., 99). The quality of fufu, however, varies with processors and season (Oyewole and Sanni, 995). The variability in fufu has been attributed to various local practices by the processors (Sanni et al. 998; Sanni, 999; Oyewole et al., 00). Fufu is traditionally sold as a wet paste, which renders it highly perishable and its poor shelf life limits its suitability for large-scale processors. A practical approach to improving the shelf life and marketability of fufu is the production of a dried product (Sanni, 00). The effect of air-drying conditions on the acidity, volatile compounds, pasting and sensory properties of fufu using Response Surface Methodology has been reported (Sanni et al., 00). Fufu samples were dried at various temperatures (5-65 o C), air velocities (- m/s) and relative humidities (0-80%). The optimum conditions for drying of fufu were 65 o C, m/s and 60%RH. These conditions reduced the concentrations of butanoic acid, but increased the concentrations of other identified volatile constituents of wet fufu. The fufu with most acceptable organoleptic properties was that dried at temperature of 65 o C, air velocity of m/s and relative humidity of 60%. Fufu has also been dried using three different drying methods namely sun, cabinet and rotary drying methods. The effect of these drying methods on the physicochemical and sensory qualities of fufu was studied. Rotary dried fufu was reported to be of good quality (Sanni and Akingbala, 000). From a recent report on the physicochemical, microbial and sensory qualities of fufu flour from various fufu processors in southwest Nigeria, there were some significant differences in the properties of fufu flour from a number of commercial and noncommercial producers (Oyewole et. al., 00). The potential for expanding the market for fufu has been reported (Dipeolu et. al., 00) and this can contribute to improving the livelihoods of cassava processors (White et. al., 00). However, there

is little information as to the consumer acceptability of fufu from a commercial point of view and how it might vary from one urban centre to another. The purpose of this study was to investigate consumer perceptions of dried fufu (commercial and experimental) and compare it with traditional forms. Majority and niche liking and socio-economic factors such as gender, age, ethnic group (tribe) were also investigated. MATERIALS AND METHODS Cassava variety Freshly harvested cassava roots (- months old; Variety TMS 0570 (is this correct not TMS 057?) obtained from the University of Agriculture, Abeokuta farm were used. Experimental samples The fufu used in this study was as follows: Traditional wet fufu The production method adopted was the typical Yoruba fufu processing method (Sanni and Akingbala, 000). Peeled cassava roots were washed and soaked in fermenting vats for days. The fermented roots were then sieved in muslin cloths. The filtrate was discarded while the sediment was collected in baskets and drained. The sediment was then packed in woven polyethylene fibre sacks and allowed to drain. Modified wet fufu Modified wet fufu method differed from the traditional process as follows: Cassava roots were mechanically grated after three days of fermentation.. The introduction of a second stage of solid state fermentation (this can be described as "double fermentation") Sieving with a reduced volume of water. Dried traditional and modified wet fufu Wet fufu slurry from traditional and modified method was dried in a locally manufactured rotary dryer (Addis Engineering Nigerian Company, Lagos). The dryer comprised an insulated drying chamber in combination with a system of duct and baffles to direct air over and through the circular chamber to promote uniform distribution (85 o C for 0min). Commercial dried fufu samples Two commercial fufu samples were purchased from Ibadan and Lagos. The Ibadan sample was dried using a fluidised bed dryer (0 o C for 5min). The method for producing the commercial sample from Lagos was not known.

Preparation of cooked fufu samples for sensory and consumer testing Fufu was cooked in boiling water (flour: water ratio of :) in an open pan with constant stirring using a wooden ladle till a strong dough was formed. (I am a bit confused here on the ratios for the wet paste i.e. the traditional product) Consumer acceptability Consumers () were interviewed at each of three locations (Abeokuta, Lagos and Ibadan) in Nigeria using the method of central location testing (Meilgaard et al. 987). During the testing, fufu samples were freshly prepared after every hour. Consumers were presented with the six cooked fufu samples at the same time. Samples (0 g) were coded with -figure random numbers and presented in random order to each consumer. Consumers were asked to score the acceptability of fufu using a 9-point verbal hedonic box scale for dislike extremely to like extremely (Meilgaard et al. 987). After testing the product, consumers were interviewed about socio-economic questions. These were about age, level of education, tribe, gender, occupation, food eaten most often, type of fufu usually consumed, how often they eat fufu, where they eat fufu, preferred foods and whether they are willing to purchase fufu flour. Sensory evaluation The same samples prepared for the consumers were selected for sensory analysis using the method of quantitative descriptive analysis (QDA) (Meilgaard et al., 987; Bainbridge et al., 996). The sensory panel (0 panellists) was conducted at the University of Agriculture (UNAAB), Abeokuta, Nigeria, under controlled temperature (air conditioned) and lighting. Sensory attributes for cooked fufu were generated during a focus group session guided by the panel leader. A total of 6 uncooked and cooked sensory attributes were developed for which the group had a consensus. Sensory attributes generated for the cooked product were dirty white colour, creamy appearance, shininess, smooth appearance, sour taste, sticky texture, soft texture, raw cassava smell. At each sensory panel, -figure coded samples (three) were served in random order. Samples (0 g) were close to room temperature and panellists rinsed their mouth with mineral water before tasting each sample. Intensity was scored on a 00 mm unstructured scale, anchored with the terms not very at the low end and very at the high end. Samples were scored in triplicate (on different days) by each panellist. Statistical analysis SPSS (0.0 Version) were used for statistical analysis in this study. Methods applied were analysis of variance (ANOVA), linear regression and principal component analysis (PCA, varimax rotation). RESULTS AND DISCUSSION Consumer questionnaire Three hundred and eleven respondents completed the questionnaire of which 6% were male and 7% female and the majority aged between 0 and 9. Most (87%) had secondary education or above and were from the Yoruba tribe. With respect to

occupation, the majority (%) were civil servants, followed by students (5%) and teachers (%); the remainder were artisans, traders and unemployed. Considering consumer attitudes to fufu, consumers in Ibadan ate fufu most often and those from Abeokuta least often. Fufu was consumed most often at home (% to 6%) while at Ibadan a higher proportion ate fufu as a street food (%). The form in which fufu was purchased varied with location. At Abeokuta, the majority (9%) purchased it in the cooked form while at Ibadan it was as flour (7%) and wet (%), while at Lagos it was in the cooked (5%) form and as flour (%). Considering eating patterns, the trend was for fufu consumption to increase and this was greatest in Ibadan and Lagos and the majority (95%) indicated that they were willing to purchase fufu flour. The preferred staples varied with location. At Abeokuta and Lagos, the preferred staples were rice followed by gari (dried fermented and fried cassava). Ibadan differed in that consumers mostly consumed yam followed by rice. Consumer acceptability Consumer acceptability varied with fufu type and location (P<0.00), gender (P=0.008) and age (P=0.050). Considering preference for fufu type (figure ), the order of liking at Abeokuta and Ibadan was similar with order from most to least liked being modified wet, traditional wet, dried traditional, commercial, dried modified and commercial. At Lagos the order of preference differed being modified wet, dried traditional, dried modified, commercial, traditional wet and lastly, commercial. These results indicate that preferences differ with location and that the dried traditional sample was liked more that the dried commercial samples. With gender, males (score of 6.0) tended to prefer fufu more than women (score of 5.6) but the difference, although significant, was not large. With age, older consumers increasingly liked the fufu (7 to 0 years - 5.7, 0 to 9 years 6. and over 50 years 6.). Fig Consumer preference for fufu samples at each location tested

Liking 9 8 7 6 5 Abeokuta Ibadan Lagos Modified wet Traditional wet Dried Traditional Commercial Dried modified Commercial Segmentation of consumer liking. Of the consumers interviewed, the individual consumer preferences were varied as illustrated in the principal component plot (accounting for 66% of the variability) in figure. The highest density of consumers were in the right-hand quadrants and their strongest preference was associated with modified wet, traditional we and dried traditional fufu. Consumers were segmented (K-means cluster analysis) into four groups with similar preferences for fufu. These segments significantly different in consumer preference and the mean likings are shown in figure and individual preferences in figure. Fig Individual variability in consumer preference

- -0.5 0 0.5 - -0.5 0 0.5 PC (%) PC (%) Commercial Modified wet Commercial Dried traditional Dried modified Traditional wet Where numbers are the relative positions of individual consumers and the segment ( to ) they were assigned to by cluster analysis, fufu samples in bold are relative positions of fufu samples. The largest group was segment (% of consumers) followed by segment (9%), segment (%) and segment (8%). Consumers in segment had a high preference for all of the fufu samples tested. Those in segment were similar to segment except that they disliked commercial sample. Consumers in segment liked the modified and traditional wet samples but had a lower preference for the others. Those in segment, the smallest (niche) group, liked all the samples except for the modified wet fufu. Fig Mean preference for each segment of consumers with similar liking 5 7 9 Dried modified Dried Traditional Modified wet Commerical Commercial Traditional wet Liking S (%) S (9%) S (%) S (8%)

The consumer segments significantly differed with gender (P=0.00), age (P=0.0), occupation (P=0.0) and where they consumed fufu. Considering gender, segments, and had a gender balance similar to the whole group of 6 males to 6 females. Segment (% if consumers) had proportionally more women (5 males to 55 females). Regarding age, segment were younger (9 years) than the other three segments ( years). The occupation of the consumers in each segment differed. Segments and were mainly civil servants (9%) or students (5 to %). Segment were students (%), civil servants (9%) or teachers (%). Those in segment were students (%) or civil servants (0%). The place where fufu was eaten varied with segments and eating fufu mostly at home (8 to 5%) or as a street food (5 to 9%) while segment was mainly at home (7%) and segment at home (5%) and at work (8%). To summarise, segments and were largely male, older, civil servants or students and ate fufu at home or as a street food. Segment comprised consumers who were mostly female, younger, mainly students and consumed fufu at home. Consumers in segment were mainly male, older, either students or civil servants and ate fufu at home or work. The consumers in each segment did not differ in terms of location, tribe, form in which fufu was purchased, level of education and how frequently they consumed fufu. Sensory analysis Analysis of variance (P<0.05) indicated that the six fufu samples were significantly different for all sensory attributes apart from sour which was excluded from further analysis. The most discriminating sensory attribute was dirty white colour (f-statistic of, 5 df), followed by sticky (f-statistic of, 5 df) and shiny appearance (fstatistic of 5, 5 df). The PCA (principal component analysis)of sensory attributes (figure ) resulted in a two-factor solution accounting for 89% of the total variation of which 7% was explained by the first PC and 7% by the second. The sensory attributes were largely separated in the direction of PC from left to right with creamy appearance, shiny and soft on the left and smooth appearance, dirty white colour, sticky and raw cassava smell on the right. In the direction of the second PC the attributes spanned from raw cassava smell to smooth appearance. The axis relating to separating the samples from left to right (PC) and up and down (PC). In the left-hand quadrants were the fufu s that were most preferred by the consumers, modified wet, traditional wet and commercial. Modified wet was positively associated with creamy appearance and inversely with raw cassava smell. Traditional wet and commercial had similar sensory properties and were positively associated with soft and shiny and inversely associated with sticky and raw cassava smell. In the right-hand quadrants were dried modified, dried traditional and commercial. Dried modified was positively associated with smooth appearance and a dirty white colour. Dried traditional and commercial were associated with raw cassava smell and negatively associated with creamy appearance. Fig Principal component plot of sensory attributes used by the panel

.5 0.5 0-0.5 Creamy appearance Shiny Soft Commercial Traditional w et Modified w et Dried modified Smooth appearance Dirty white Sticky Raw cassava smell - -.5 Commercial Dried Traditional - - -.5 - -0.5 0 0.5.5 PC (7 %) Internal preference mapping Internal preference mapping of the individual preference ratings revealed differences in the preferences among consumers (figure 5). The segment means are shown Fig 5 Internal preference map for Nigerian fufu.5 PC (%) 0.5 0-0.5 - Dried traditional Dried Modified Sticky S Dirty w hite Smooth S Raw cassava smell Commercial Traditional wet Shiny Commercial S Soft Creamy S Modified wet -.5 -.5 - -0.5 0 0.5.5 PC (%) where: S, S, S amd S are location of consumer segment means. Models to relate consumer preference to sensory scores Regression on mean consumer preference for each segment and mean scores for the sensory attributes

S creamy The failure to correlate the mean sensory response with the mean preference response for each consumer segment suggests that factors other than the sensory attributes used in this study are contributing to fufu preference. These might be concerned with packaging for example. Individual consumer preference and mean scores for the sensory attributes External preference mapping has been proposed as a tool for modeling the relationship between consumer and sensory data. However, the limited number of samples used in the central location tests negates its application. Therefore, an alternative approach using linear correlation coefficients of individual consumers with the mean sensory attributes was investigated. This was used to determine if there were groups of consumers who might correlate with the sensory panel scores enabling models to be developed for sub groups of the consumers. Having obtained a table of correlation coefficinets, hierarchical cluster analysis (Wards Method) determined four groups of consumers that had similar linear correlation profiles with respect to the preference and sensory attributes. The mean of the correlations for each group are given in figure 6. Fig 6 Mean of linear correlations (individual consumer preference with mean sensory scores) for consumer groups identified by cluster analysis.000 0.500 Mean correlation 0.000-0.500 Dirty white Shiny Creamy appearance Smooth appearance Sticky Soft Raw cassava smell S (%) S (%) S (%) S (%) -.000 Of the groups, the consumers segments and (% of consumers) followed similar patterns expect that segment had higher mean correaltion coefficients. Consumers in segments and (66% of consumers) had mean correlation coefficients that had opposite signs to segments and. The preference scores of groups and and and were combined and correlated against the sensory scores. Consumer preference of groups and significantly

correlated with dirty white colour (cc 0.96, P = 0.00), sticky texture (cc 0.907, P = 0.0), raw cassava smell (cc 0.87, P = 0.08) and sour taste (cc 0.6, P = 0.0) and groups and significantly correlated with creamy appearance (cc 0.886, P = 0.09). Conclusions Acknowledgements References

Table Social and economic characteristics of fufu consumers in Nigeria Location Abeokut a Ibada n Lago s Total Gender Male 60 6 67 6 Female 0 6 6 Age 0-9 0 0-9 5 5 9 0-9 8 0 7 0-9 9 9 7 50-59 7 5 5 60-69 0 0 0 Education No formal education 7 0 7 8 Primary School 7 0 0 Secondary school 9 5 6 OND/NCE 0 HND/BSc 55 6 Occupation Civil service 9 0 7 Student 5 8 7 Teaching 6 8 Trading 0 5 7 7 Artisanship 8 Unemployed 0 0 0 Food usually consumed Rice 7 0 7 8 Gari (fermented and fried cassava) 8 5 8 Beans 9 8 Yam 8 Fufu (fermented cassava) 5 0 Pounded yam 8 0 7 9 Dodo (fried plantain) 8 Amala (dish made from dried yams) 0 Lafun (sun-fried cassava) 0 Tuwo (stiff porridge made from sorghum flour) 0 0 Frequency of eating fufu Never Occasionally 5 7 9 6 Once a month 5 Once a week 6

Everyday 7 Where consumers eat fufu most often Home 57 6 55 Street food 9 Home and street food 0 8 Home and workplace 8 6 9 Workplace 6 6 8 Form in which fufu is purchased Cooked 9 5 50 Flour 7 Wet 9 6 Wet and cooked 9 5 9 Cooked and flour Wet and flour 0 0 Change in amount of fufu eaten in last year No change 5 7 Increased 8 7 6 Decreased 9 Has consumer heard of fufu flour? No 55 6 5 Yes 59 5 7 6 Is consumer willing to purchase? Yes 96 9 96 95 No 8 5

Percentage Distribution Characteristics Abeokuta (n = 99) Lagos (n = ) Ibadan (n = 9) Total (n = ) Sex Male Female 59.6 0. 66.9. 6.7 6. 6. 6.6 Age (years) 0-9 0-9 0-9 0-9 50-59 60-69 0.00 5.5 8...0 0.0.7.. 9.0 5.0 0.0. 5.7 9.8 8.7 6.6.. 8.8 7.0 7. 5. 0. Education No formal Primary Secondary OND/NCE HND/BSc. 7. 7. 9... 6.6.6.. 6. 0.0 9.9 5. 9.9 5.9 7.9 9.5 5.9. 5. Occupation Civil service Teaching Artisanship Trading Student Unemployed 9. 6..0 0.0 50.5 0.0 7.7.5 0.8 6.6 7. 0.0 9.8 7.6 7.7 5. 8.5. 0.6.. 7. 5.5 0. Tribe Yoruba Hausa Igbo 9.9.0.0 80. 0.8 9.0 8.5.. 86. 5.7 8.07