Utilization of Pepeta, a locally processed immature rice-based food product, to promote food security in Tanzania

Consumption of pepeta, a traditionally processed immature rice product, is common in Tanzania, where it contributes to food security as an early rice food i.e. when there is no other food available at the household while the crops in the field not yet fully ripe. Many production steps are needed to guarantee a consistent, good quality pepeta product, and this hinders its utilization in other rice-growing sub-Saharan regions. This study aims to gain insight into the pepeta processing knowledge and final product, and assess variations in the processing conditions and parameters across the study area. A survey among 257 Tanzanian processors and consumers revealed that the pepeta product is widely known, rated second (73.5% respondents) as rice-based food after wali (cooked white rice, (100%)) and linked to traditions of the communities in the study area. Harvest of immature rice grain, roasting, pounding, cleaning, and packing are the main process steps of pepeta production. Method of rice harvest, rice suitability for pepeta production after optimum harvest, dryness of grains and number of pounding as indicator to terminate roasting and pounding process respectively, and packaging materials used varied significantly across respondents in the study area. Reported criteria considered by respondents for product acceptability did not vary significantly across study area. The criteria include colour (76.5%), general appearance (60.8%), texture (64.7%) and taste (52.9%). Immature rice paddy and pepeta were sold at a higher price than mature rice paddy and white rice, respectively, which implies that options to facilitate pepeta processing through, for instance, standardization of processing conditions and parameters could lead to increased income.


Introduction
Rice (Oryza sativa L.) is an important staple crop for global food security. Its global utilization is estimated at 503.9 million tonnes (milled basis), of which 80.5% in food uses accounting to a per capita food consumption of 53.9 kg [1]. In the sub-Saharan region, including Tanzania, chain in the study area were investigated as well. The documented information serves as a basis for research into possible ways to optimise specific processing conditions to improve nutritional quality, and/or other fields along pepeta value chain to enhance its competitiveness.

Study design
Both quantitative and qualitative approaches were used in this study. A face-to-face interview through questionnaires together with focus group discussion and on-field observation via demonstrations were the main data collection tools used. The questionnaire was based on a conceptual model that represent simplified pepeta value chain and its link to food and livelihood security.

Questionnaire design
2.2.1 Conceptual model used to design the questionnaire. Fig 1 show a conceptual model used to design the questionnaire (S1 Questionnaire) and checklist (S1 Checklist). The model reflects the key actors (rice farmers, processors and consumers) along pepeta value chain and its linkage to food and livelihood security. Immature rice grains are sole ingredients for processing of pepeta products, the availability and accessibility can substantially affect pepeta processing and consumption. Rice is preferred for human consumption and plays a significant role in Tanzanians' culture, traditions, and religion [22]. It is a major source of income, food and employment in rural areas, providing about 95% of the national food requirements, accounts for more than 70% livelihoods of the Tanzanian population [23]. The influence of processing conditions on product quality properties and acceptability are fundamental in processed food products like pepeta [24], affecting food product utilization.

PLOS ONE
Therefore, indigenous knowledge on immature rice grains harvesting and processing, information on pepeta product quality and acceptability, and existing problems are included in the model as well. Food product characteristics such as appearance, taste, texture, colour, aroma, and socio-economic characteristics of consumer like gender, age, education and income can influence food choices and preference [25].
2.2.2 Questionnaire. The questionnaire was developed based on the conceptual model (Fig 1) to collect information on the following aspects: (1) General information and awareness of pepeta, (2) Pepeta processing knowledge, (3) Product quality criteria and utilization, and (4) Product processing, marketing and storage constraints. It was divided into sections containing specific questions to extract information from different actors along pepeta processing value chain (S1 Questionnaire), as follows: • General information: Gender, age, education, income source, awareness about pepeta and type of actor in the value chain, i.e., processor, consumer and/or trader.
• At consumer level: Consumer information on consumption frequency of pepeta. Place of purchase (e.g., informal market, farmer-gate, and/or supermarket), prices relative to product quality (freshness and seasonality). Properties considered during purchase (e.g. colour, appearance, texture, taste and aroma).
• At processor level: Pepeta processing techniques and product quality. Sources of raw materials (immature paddy), parameters considered when purchasing raw material for pepeta processing (i.e., varieties and maturity level) and prices. Places of sales of pepeta, consumer perception of product quality (e.g., colour, appearance, texture, taste and aroma), constraints and most tedious process steps.
• At trade level: Places of sales of pepeta, consumer perception of product quality (e.g. colour, appearance, texture, taste and aroma).

Study area
The study was conducted in the Eastern and Southern Highland parts of Tanzania [26] The district administrative areas were chosen based on the relevance of rice as one of key crops in the farming system and its dual purpose as a major source of households' income and food security. According to the perception of local people (community leaders) and Village Agriculture and Extension Officers (VAEOs), the area has also been serving as a centre of pepeta processing knowledge for many decades.

Sampling of respondents/informants
Prior to data collection, a random observation was done on 117 food venders selling food at local markets to estimate the proportion "p" of venders selling rice-based food products. The research team assessed whether a vender was selling at least one common rice-based food product found in the community. The obtained proportion was then used to estimate the total number "n" of respondents to be interviewed in the study, using the following formula [27]: where n is the total sample size; p is the proportion of targeted informants; U 1−//2 = value of the Normal random variable for a probability value of 0.975 (or α = 0.05), U 1−//2 �1.96; d = margin error of the estimation fixed at 5% (0.05). The district population size was used to compute the number of participants to be interviewed for each district council (Table 1) according to [28]: where n i is the number of participants in the district council; N i is the total number of the population in the district council; and N is the total number of people living in the five selected district councils. The used population data were based on Tanzania National Bureau of Statistics report [29].

Field data collection
Data collection was from September 2017 to February 2018, using a structured questionnaire, by focus group discussions and observations. The questionnaire was first tested and adjusted before being administered to the 257 respondents selected from five district administrative boundaries (Table 2). For face-to-face interviews, respondents from each household in the selected villages were randomly selected from the list of rice farmers and/or pepeta processors who were willing to participate provided by VAEOs. The interviews were held in Swahili, the national language of Tanzania, which was well understood by all respondents. Focus group discussions were employed to comprehend the collected information from individual face-to-face interviews, as people normally tend to mention the most important things when asked to freely recall under a given short time [27]. Such discussions consisted of 12-15 participants, purposively selected following criteria such as age (youth, people of reproductive age and old), gender balance, and different actors (farmers/consumers, processors and traders) along the pepeta processing value chain ( Table 3). The focus group discussions last for 2-3 hours for each conversation and were guided by a checklist. A total of 5 focus group discussions were conducted; 3 in Kilombero district and 2 in Ulanga district, depending on the population sample size. The observation data were collected during harvesting and handling of rice and pepeta processing to verify data collected from interviews. Physical properties data i.e. moisture content, weight, temperature and duration of various processing steps of pepeta production were measured using a digital grain moisture meter (SATAKE, MOISTEX SS7), weighing scale (Endel™, EWS-H-PLUS), digital thermometer (Fluke, model 52 II), and digital timer (Fisherbrand™), respectively. The study was approved by and conducted in collaboration with Tanzania Agricultural Research Institute (TARI)-Ifakara center, following all relevant regulations. The individual pictured in Figs 5 and 6 has provided written informed consent (as outlined in PLOS consent form) to publish their image alongside the manuscript.

Data processing and analysis
Data from individual household interviews were subjected to descriptive statistics (percentages or frequencies) using IBM SPSS statistics (version 23, USA) and Microsoft Excel 2016. A chisquare test of independence at 0.05 level of significance was performed to determine if there was significant different between pepeta processing practices of Kilombero and Ulanga respondents and ascertain the relationship between pepeta knowledge (awareness) and respondents' demographic data (gender and age). The chi-square test was performed independently for each individual option in the multiple response questions. Data for process efficiency, and moisture, weight and heat losses, the independent t-test (p < 0.05) was computed to evaluate any significant different between Kilombero and Ulanga districts. Qualitative data gathered through focus group discussion and observation were content analysed. The collected GPS data were used to prepare a map using QGIS (version 3.4.3). Pepeta processors/traders 6 7 Total (per FGD) 12 15 Number of FGD conducted 3 2

Product history, taboo and uses
According to the focus group discussions, pepeta means "flattened grains" and its processing knowledge dates back since introduction of rice in the community. This knowledge has been passed on from generation to generation. Pepeta and its processing knowledge form an integral part of social rites, rituals and festivals of the Ndamba, Mbunga, Ngindo, Pogoro, Kwere and Doe ethnic tribes found in the study areas as mentioned by respondents. According to the interviewees, no girl would qualify for marriage if she would not know how to process pepeta. Therefore, pepeta processing was one of the trainings given to young girls during unyago rituals, a practice to celebrate the coming of age of girls or during weddings. According to kwere and doe tribes, a grounded pepeta product, known as bwimbwi, was used as a participation fee in jando, a circumcision rite for boys. Unyago and jando rituals involved instructing youth about sex and conjugal life [30]. It was further believed that a marriage became happy if a spouse (wife) prepared a large amount of pepeta to proudly serve her husband throughout a year or until the next rice harvesting season.
Preparing pepeta is a way of showing affection and therefore pepeta was also given to special guests like in-laws as a symbol of respect and care. Reserving traditional snacks for special, esteemed individuals like men and in-laws is a common practice among Tanzania communities [31]. Pepeta was also recognized as a symbol for the start of a new harvest season for Ndamba, Mbunga, Ngindo and Pogoro tribes. No household was allowed to start the new harvest before pepeta was prepared and sent to the chief, a community leader, for making a ritual sacrifice asking protection against natural calamities, wild animals and birds.
In the study area, respondents explained their reason for using immature rice to prepare pepeta: as a means of securing food when no other food was available at the household while the crops in the field not yet fully ripe. Therefore, mothers were forced to prepare pepeta to feed their children while waiting for the main dish. These findings illustrate the tremendous importance of pepeta in the community living in the surveyed areas: "a means of communication, affirming and reinforcing social relations, of expressing one's personal or group identity and of connecting to the living or ancestral peer group [32]".

Product popularity
Though the popularity of pepeta varied significantly (x 2 = 109.193, df = 4, p = 0.000) across respondents in the study areas, most of the respondents (73.5%) knew the product and rated the product second after wali (cooked white rice) (Table 4), the main form of consuming rice in Tanzania [23]. Pepeta was mentioned as one of the common rice-based food products by 100% of respondents in Kilombero and Ulanga districts, and by 79.2 and 70.8% in Bagamoyo and Mvomero rural districts, respectively. Contrarily, only 23.6% of respondents from Iringa rural districts knew pepeta. This difference could be due to observation that pepeta was no longer processed (hence unavailable) in Iringa rural district, whereas it is regularly processed for both sales and household consumption in Kilombero and Ulanga districts, and occasionally processed in small quantity for household consumption in Mvomero and Bagamoyo districts. Pepeta processing, as traditional knowledge, is attributed to several factors including gender and generation [32]. To confirm this concept, the association between pepeta knowledge (awareness) and respondents' demographic data (gender and age) among districts (Mvomero, Bagamoyo and Iringa rural) was tested. The analysis in Table 4 show gender differed significantly (x 2 = 8.817, df = 2, p = 0.012) across respondents in the districts, more males knew pepeta product in Bagamoyo and Iringa rural districts, and the vice versa is true for Mvomero district. However, no significant differences were found in the overall respondents between males (53.0%) and females (47.0%) for pepeta knowledge (x 2 = 3.417, df = 1, p = 0.065), indicating location is the main influencing factor for the observed differences across districts in the study area. To evaluate the impact of age on pepeta knowledge, respondents were categorized into three groups: below 30 years (regarded as youth with little or no knowledge of pepeta), between 30 and 44 years (regarded as adults with moderate pepeta knowledge), and 45 years and above (considered as elders with much experience on pepeta). There was a significant difference in the overall respondents regarding pepeta knowledge among age groups (x 2 = 9.696, df = 2, p = 0.021), with 61.5% of those familiar with pepeta (55% of the respondents) being 45 years and above, while 12% were below 30 years (Table 4). Also, age groups differ significantly (x 2 = 6.265, df = 4, p = 0.018) across the respondents in the study area. The data suggest that pepeta processing, like other traditional knowledge, is at risk of extinction due to adaptation to surroundings and culture changes from one generation to another [32]. In this study no data about pepeta processing and/or consumption was collected at Mvomero, Bagamoyo and Iringa rural districts as respondents in these districts could not recall the last time they processed and/or ate pepeta. They obtained pepeta from local markets or as a gift from Kilombero and Ulanga folk.

Processing season and availability
Pepeta is processed in large amounts twice a year, following the rice production calendar, which controls the availability of immature rice grains. The main production season is from April to July, i.e., the major rainy season when immature rice is sourced from both rainfed and irrigated fields, and the second season is from October to December when irrigated fields are the only source of rice. Pepeta processing as a source of household income was only mentioned by 50.0% (Kilombero) and 52.6% (Ulanga) of the respondents, the informants in the survey (Table 2), and was exclusively an activity of women in the visited areas. These findings are in line with previous research [33,34], which underpins the important role women play in preserving and transferring traditional food processing knowledge from generation to generation in sub-Saharan region.

Pepeta processing
Pepeta resembles flaked rice, but the use of freshly harvested immature rice grains make the product and production process distinct from many traditional and commercial flaked rice products in Asian and other rice-consuming countries [35]. The main processes involved in the production of pepeta are harvesting of immature rice grains, cold soaking, roasting, pounding, cleaning and packing (Fig 3). Though the production is mainly at household level, variations exist for pepeta production destined for sale compared to that intended for home consumption, mainly regarding processing quantity. For commercial production, 60-180 kg of pepeta per interviewed processor and less than 15 kg of pepeta per household for home consumption is processed during each cropping season. During pepeta processing, rice grains undergo several physical changes, including drying, a colour change from greenish yellow to bright yellow, dehusking and flattening (Fig 3A-3D). Table 5 shows the moisture content, duration, weight and temperature for various processing steps of pepeta. Selection of rice cultivar and optimum harvest time are critical steps for the quality and flavour characteristics of pepeta. Generally, pepeta is prepared from different aromatic landraces of rice (Table 6). Each informant cited at least four cultivars known to them and/or found in their locality. The most frequently cited cultivar was TXD306 mentioned by 100% and 95% of respondents in Kilombero and Ulanga districts, respectively. Participants were also asked to specify the most preferred cultivar. TXD306 (87.5% of respondents), Kalimata (7.1%) and Mbawambili (5.4%) were most preferred (Fig 4), and differ significantly (x 2 = 7.554, df = 2, p = 0.023) between Kilombero and Ulanga respondents. Respondents considered TXD306, which is a semi-aromatic hybrid variety, to have quality traits and availability advantages over highly-aromatic landraces with a long growth cycle (i.e., Kalimata and Mbawambili). TXD306 can be cultivated twice annually, guarantees the availability of immature rice grains and hence pepeta processing. In addition, the awns of the Mbawambili variety tend to burn during roasting, introducing black spots and thereby affecting the general appearance of the pepeta product.
Harvesting of immature rice grains. Harvesting immature rice destined for pepeta processing involves cutting, threshing and cleaning. Rice in the field is harvested by cutting rice panicles (60%) using a snail shell or cutting rice stalks (40%) using a sickle ( Table 7, Fig 5A and  5B), and differed significantly (x 2 = 16.042, df = 1, p = 0.000) between respondents of Kilombero and Ulanga districts. Harvesting by cutting rice panicles involves sorting of appropriate rice grains, leaving rice in the field that is considered unfit (too immature or too mature) for pepeta processing by processors. In contrast, the entire field is harvested when using sickles. Changes of leaf and panicle colour (92.7% of respondents) as rice matures, and biting through grain (25.5%) are common maturity indicators for optimum maturity of rice in the community, harvested two weeks (50.9% of respondents) or three weeks (32.7%) after 50% flowering of rice in the field. The colour of the grain seed coat changes from green to a distinct colour in accordance with rice cultivar at the onset of ripening, and grain endosperm hardens as it develops [37,38]. Though no significant different in the indicators for optimum rice maturity, the suitability of rice in the field after optimum maturity varied significantly (x 2 = 17.595, df = 2, p = 0.000) between respondents of Kilombero and Ulanga districts. According to respondents, rice in the field remained suitable for one (63.6% of respondents) to two weeks (34.5%) after attaining the optimum maturity level for pepeta processing.
Threshing, the process that involves separating rice kernels from panicles but not removing the husk, is done immediately after cutting by hand (Fig 4C) or by beating with a stick on the paddy to maintain freshness of the harvested rice kernels. After threshing, rice grains are cleaned by a combination of winnowing and hand sorting (Fig 4D). Unfilled kernels (92.7% of informants), rice straws and chaffs (81.8%), stones (36.4%) and weed seeds (27.3%) were major unwanted materials removed during cleaning ( Table 7). The unwanted materials removed during cleaning did not differ significantly between Kilombero and Ulanga respondents, except for weed seeds (x 2 = 11.786, df = 1, p = 0.001). The moisture content of fresh harvested rice kernels ranged from 30-36% (Table 5).
Soaking. We observed cold water paddy soaking as an optional process, mainly in commercial production. It is done when harvesting until late in the evening to prolong freshness of harvested paddy until the next day. Table 7 indicates 67.3% of the respondents soaked freshly harvested immature paddy overnight, typically into 10-20 litre plastic containers. Paddy soaking is also said to soften and standardize the moisture content of harvested paddy kernels. Indeed, rice in the field matures heterogeneously, and large variations up to 46% in individual kernel moisture content at harvest have been reported [39,40]. Generally, freshly harvested immature paddy is preferred, as it gives a more white-greenish colour. The soaked paddy is water drained at ambient temperature for about 2-4 hours before roasting.
Roasting. In the study areas, fresh or soaked paddy is roasted in dry aluminium ( Fig 6A) or earthenware (Fig 6B) pots, with continuous stirring using a big wooden spoon. We noticed no roasting medium such as sand or fine silt [35] used in many related flaked rice products. About 0.25 to 0.41 kg of paddy was roasted at a time for about 3-8 min. The roasting vessel was warmed up to 117-180˚C and paddy reached a temperature from 80-129˚C (Table 5). Table 8 show many of the respondents mentioned 1-5 min (49.1%) and 6-10 min (29.1%) as roasting duration. However, 56.4% of the informants indicated variations in the duration of roasting, depending on the amount of paddy roasted at a time, hotness of the firewood flame and efficiency of stirring.
Pepeta processors depend on knowledge, experience and observations to determine when to terminate the roasting process. Dryness of the grains (90.9% of respondents), puffing of grains (56.4%) and colour changes of grains (52.7%) were common indicators to determine the end of the roasting process (Table 8, Fig 3B). There was no significance difference on various roasting parameters between Kilombero and Ulanga respondents, except for dryness of the grains (x 2 = 4.526, df = 1, p = 0.033) a factor used as indicator for termination of roasting process. According to processors, dryness of the grain could be assessed by the ease of stirring due to a decrease in adhesion as grain dried during roasting, and/or easiness of dehulling when roasted grain is rubbed between hands. These findings differ from [35], who reported the initiation of a popping sound (puffing of grains) as the main indicator used for the termination of roasting in preparation of rice flakes.
Pounding. Pounding is the labour intensive and critical operation of pepeta processing, traditionally done by hand pounding the roasted paddy using a pestle and mortar (Fig 5C). Through pounding, the roasted paddy gets dehusked and flattened concurrently. Hand pounding is common practice when preparing traditional flaked rice from mature rice at household level in India [35]. Many respondents (65.5%) mentioned variations in the duration of pounding, where 61.8% and 21.8% reported a pounding duration of 1-5 min and 6-10 min, respectively (Table 8). However, during various pepeta processing demonstrations, about 0.17-0.34 kg of roasted paddy was pounded for 1-3 min by two or three people at a time (Table 5). We observed that pounding speed and hotness of the roasted paddy were important parameters. For effective processing, the pounding should immediately start and end while the Table 7. Overview harvesting of immature rice grains and its preparation before pepeta processing.

PLOS ONE
roasted paddies are still hot. Even a slight delay from pounding may affect the thickness of the flattened grains and general appearance of the end product. Absence of undehusked paddy grains (92.7%) and flatness of the pounded grains (70.9%) were considered by respondents as major indicators for termination of the pounding process (Table 8, Fig 3C). No significant difference observed between Kilombero and Ulanga respondents in pounding parameters, except for number of pounding (x 2 = 6.149, df = 1, p = 0.013) used as a factor to end the pounding process, mentioned by respondents (25.7%) in Kilombero district only. Cleaning and storage. Fig 5D shows the cleaning process using the traditional winnowing method. The pounded grains are cleaned to remove by-products of the pounding process, namely broken grains (74.5% of respondents), husk/brans (87.3%), and undehusked grains (72.7%) ( Table 8). In this study we found that the cleaning was done in two stages: first cleaning immediately after each pounding to remove brans, and a second cleaning at the end of pepeta production to remove remaining brans, broken and undehusked grains (Fig 3D). According to processors, any delay to remove brans immediately after pounding, affect whiteness and hence general appearance of the pepeta product. After cleaning, pepeta is stored in polyethylene bags (62.7% of respondents) and plastic containers (31.4%) of different sizes. Aluminium pots, earthenware pots and paper bags are seldom used. Although there was no significant difference in cleaning parameters, the packaging material used i.e. plastic containers (x 2 = 4.483, df = 1, p = 0.034) and paper bags (x 2 = 5.175, df = 1, p = 0.023) differed significantly between Kilombero and Ulanga respondents, with Kilombero being more sustainable-oriented. Paper bags which were only mentioned by respondents in Kilombero (24.2%), are considered as eco-friendly sustainable packaging material due to its biodegradability properties Table 8. Overview of main pepeta processing steps in the study area.

Process efficiency and losses
Several terms are used by experts to evaluate the efficiency of processing methods. In this study we use pepeta recovery percentage to evaluate the process efficiency of pepeta processing. Based on the rice milling recovery concept [42], we define pepeta recovery (i.e. process efficiency) as the percentage of pepeta yield based on the initial paddy weight before roasting. The survey found 40% recovery when immature paddy was processed into pepeta (Fig 7), a lower value than the 50-60% [43] when mature dried paddy is milled. The apparent low yield of pepeta attributed to 35% weight loss due to remove of unwanted materials (husk, bran dust, broken and undehusked paddy grains) during cleaning, 20% weight loss during roasting, and 6% weight loss due to grain scattering during pounding. High moisture loss (65%) was the main factor contributing to weight loss during roasting. However, processing efficiency, loss at roasting and cleaning as well as heat loss from flame to paddy differed significantly between Kilombero and Ulanga respondents, with Ulanga being more efficiency compared to Kilombero.

Pepeta trade
In the surveyed area, pepeta trade is largely carried out in the Morogoro region, with by far the largest proportion of the commercial production and distribution chain dominated by women. The trade starts in Kilombero and Ulanga districts, which are the centres for pepeta production in the region. In these districts, processors bought immature paddy directly from the field, which was sold at a price almost three times that of dried mature paddy. This could be a way to offset losses due to the fact that when harvesting immature rice the yield is lower than for mature rice [44]. According to respondents, prices varied from one place to another, and depending on seasonal availability: lower during bumper harvest and higher during scarce supply. Pepeta was sold mainly along roadside selling centres (100% of respondents), seldomly at local markets (18.2%) and train stations (10.9%) ( Table 9). Although there was no significant difference in roadside selling centres, pepeta selling at train stations differed significantly (x 2 = 3.848, df = 1, p = 0.049) between Kilombero (17.1%) and Ulanga (0.0%) respondents. It is important to note that Ulanga district is not accessible by train as Kilombero district. The study found three major pepeta roadside selling centres, namely  E) in Ulanga district. The processors who are also traders, have formed their groups, which regulate the number of processors selling the product in a daily rotation routine. This is because there are more processors than the capacity of the vending centre. However, processors complained about a lack of reliable markets opportunities, indicating the challenge of pepeta distribution chain and the potential to improve the trading network. The price of pepeta did not change during the processing season and was twice that of milled white rice. This is illustrative of the importance of adding value to agricultural produce for household income generation and hence livelihood and food security improvement [2,45] This study found that pepeta was also sold to the neighbouring municipalities such us Morogoro and Dar es Salaam by middle-persons (3.6% of respondents). Processors mentioned to produce pepeta for either selling (50.9%) or both selling and household consumption (49.1%). Though the contribution to household income has not yet been documented, most processors cited that it covers by far the greatest proportion of their daily needs such as food, clothes and school fees. Fig 8 indicates factors considered by respondents in the study area when buying and consuming pepeta. According to respondents, major factors considered were colour (76.5% of Table 9. Overview of pepeta trade in the study area.

Kilombero (n = 35) Ulanga (n = 20)
Pepeta uses respondents), general appearance (60.8%), texture (64.7%) and taste (52.9%). In addition, aroma, flatness and number of broken grains were also considered. Location had no effect on quality criteria considered by respondents between Kilombero and Ulanga district. Colour was mentioned as a highly critical factor in the focus group discussions as well. According to the interviewees, a white-greenish colour (Fig 3D) is most preferred as it indicates that the product is made from immature rice grains. The informants cited general appearance to include presence or absence of black spots, husks, bran, undehusked grain, dust, broken, and flatness of pounded grains. We observed that the general appearance and colour of pepeta product act as eye-catching traits, highly differentiate pepeta products among processors and play a key role in consumer decisions to taste the product before buying. This is possible because pepeta is marketed in unpacked form, whereby consumers are invited to taste the product before buying. After tasting, flavour and texture played decisive roles in consumers' buying decisions. According to interviewees, a good pepeta texture is not too crunchy, moderately soft but not too sticky and thin. Thick and hard pepeta was reported to cause jaw aching even when consumed only for a short time. Generally, the use of immature rice grains gives a butter-like popcorn flavour to pepeta, which is highly appreciated by many consumers (98%) in the surveyed areas. Indeed, the maturity level of cereal grain has been associated to the taste of Firiks (or Frekeh), a traditional product prepared from immature wheat [16,46]. We found that the most common way of consuming pepeta is as a snack (98.2% of respondents), whereas 1.8% used pepeta as breakfast. In this study, snack refers to consumption as a leisure activity rather than as a meal, with no specific time. Fig 9 presents a general overview of pepeta processing and quality problems identified by respondents in the study area. Pepeta processors used colour changes of rice grains and hardening of endosperm as criteria to identify the optimum rice maturity level for pepeta processing. This method is very subjective, resulting in variations in maturity levels of the harvested rice grains, and hence colour and flavour of pepeta end products among processors.

Pepeta production problems
Harvesting by cutting and threshing immature rice grains is done under wet conditions, a period when water in the rice fields is still needed for crop growth. Generally, it is difficult to control the water level within individual rice fields due to poor infrastructure. This makes the use of local harvesting tools and methods inevitable as mechanised tools cannot be used in wet conditions, making the process laborious and time-consuming. In some occasions, processors soak immature paddy in cold water to extend the freshness of harvested paddy until the next day when harvesting ends late in the evening. However, a focus group interviewee said that soaking imparted undesirable colour changes and off-flavour to the end product, similar complaints as reported by [35].
Dryness, puffing and colour changes of rice grains are important factors used by processors to determine the end of the roasting process, a subjective method based on past experience and visual observations. In addition, these factors highly depend on the amount of paddy being roasted, temperature and stirring speed. Though the amount of paddy is easily controlled by roasting a known amount, maintaining a constant temperature and stirring speed is difficult due to the use of firewood and manual stirring. Properties such as wood density, moisture content, flammability, flame brightness, and flaming period affect flame temperature [47]. These are not considered much by processors as the type of firewood used in the study area is based on availability. Usually, roasting is done in open air, a situation which further aggravates the problem of uncontrolled heat distribution during roasting due to air velocity. An uncontrolled heat flow and inefficient stirring result in over-roasted paddy due to high heat or under-roasted paddy due to false puffing of rice grain, respectively. This causes a poor quality of the end product, with defects concerning moisture content, softness, colour and flavour (Fig 7). To prevent such issues, processors have to continuously stir while regulating the amount of chopped wood to maintain a proper heat flow during roasting, a tiresome process according to processors.
About 72.7% of the respondents mentioned pounding as the most tedious process step, at least two times more frequent than the remaining processing steps: threshing (23.6%), cutting (18.2%) and roasting (10.9%) ( Table 8). Application of labour serving technology like roaster and rice flaking machine [48] will substantially relief the workload. To achieve the desired flatness, processors have to manually pound roasted paddy at a high speed to finish the process quickly while the paddies are still hot. This is because cold pounded rice grains do not flatten; instead normal milled white rice kernels are obtained. Processors complained about physical discomfort, including hand, back and chest pain.
Respondents mentioned an impairment of whiteness and general appearance of the end product when cleaning is delayed after pounding. Other constraints include quality degradation due to moisture uptake by the product, and contamination by insects, dust and mould as a result of poor storage facilities and hygienic conditions. Pepeta as a dried product, tends to absorb water from the surroundings. To maintain product quality, pepeta was stored in plastic containers with tight-closing lids and kept in cool and dry conditions. In addition, respondents in the focus group discussion mentioned to re-sundry pepeta to extend its shelf life when stored for a long period, especially during the rainy season. The results concur with [33], who reported re-sundrying of traditionally prepared dry monkey orange (Strychnos spp.) products to extend shelf life. Wet and high humidity storage conditions facilitate clump formation, mould growth, discolouration, a bad smell and bitter taste; factors that were much considered by consumers when rejecting a pepeta consignment.

General utilization of immature cereal-based products
Cereals are the major staple food for many people. Wheat, maize and rice comprise at least 75% of the world's grain production [49]. Consumption of immature maize as a roasted product or boiled whole kernel, and/or traditional processed products is common in Africa [50,51]. Mohlefe or malitsibana is a common green maize bread in Lesotho, prepared by wet-milling maize kernels to a thick dough, shaped into cob-like forms, covered with maize leaves and steamed until completely cooked [51]. Similar to pepeta, the preparation of mohlefe is done during the harvest season when households are waiting for crops to mature in the field. Degree of maturity and varietal differences significantly impact the quality properties of mohlefe. Preferred attributes are a whitish or yellowish colour, depending on the maize variety used, and an intense aroma of immature green maize kernels [51]. Contrary to pepeta, salt is added during preparation of mohlefe, and this food is mostly consumed as breakfast.
Firik (also known as frikeh or frekeh or freekah) is a common immature whole wheat-based food consumed in the Middle East and North African countries [16,52]. Processing knowledge and quality attributes of firik have been extensively studied [16,46,[53][54][55]. To prepare firik, immature wheat ears are scorched or roasted on open fire, sundried, threshed, after which the kernels are separated from the hulls and cracked [46]. Similar to pepeta, harvesting time, wheat cultivar and processing conditions determine the quality attributes of firik. The best harvest time for firik ranges from the late-milk to mid-dough stages, which gives a better taste than for the ones processed at the full ripe stage. Hard durum wheat (Triticum durum) is preferred for firik production. Scorching gives the firik its unique, appetizing smocked flavor [16]. Generally accepted high quality firik is plump, firm when fresh, slightly burnt, green when dried, containing few remains of the pleas, lemmas, glumes, and free from stones and debris [52]. Firik is widely used as an ingredient in the preparation of some specific meals, and especially consumed with meat, tomatoes, stuffed squash, eggplant, grape leaves, and chicken broth [16,46].
Novel processing of immature grains to increase their applications in food industry has been documented. Previous study [56] assessed the application of fluidized-bed coating technology in improving the health benefits i.e. antioxidant activities of Khao Mao cereal product (a traditional puffed pounded-unripe rice in Thailand). Yilmaz et al. [18] investigated the potential use of infrared radiation to unfold the limitation use of immature rice grains due to high rancidity rate compared to mature rice grains. Infrared stabilized immature rice grain flour has been used in preparation of extruded rice products [57], gluten free bread [17], and Tarhana, a cereal based fermented food [58]. Ç etin-Babaoğlu et al. [59] and Pepe et al. [60] evaluated the application of immature wheat flour in straight-dough and sour-dough (fermentation) processes for production of wheat bread. In addition, fermentation technology has been widely used to improve the nutritional quality of other immature cereal-based products like green spelt wheat tempe [61], and green maize (melie) and green sorghum (senkhoana) breads [51].

Conclusions and recommendations
This study presents the current use, potential and challenges of pepeta and its traditional processing knowledge on promoting physical and economic accessibility to early season food, which are main aspects of food security, at household level in Tanzania: "Food security is achieved when all people, at all times, have physical and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life" [14]. Selling immature rice and pepeta provides economic benefits for local farmers and processors since these products are commonly sold at prices that are about three times higher than for mature paddy and white rice. The premium price earned contributes to the livelihood of the household.
However, the current traditional processing method has disadvantages: the process is slow, laborious and not suited for upscaling as this cannot guarantee a consistently good quality pepeta product. There are no standards for rice harvesting, and processing conditions and parameters. These problems affect its full utilization across sub-Saharan Africa, further aggravate food insecurity in the region. In order to develop a nutritious high sensorial quality pepeta product with better shelf life and ease of portability, research on the factors which influence its quality including nutritional, physico-chemical, rheological and aroma properties is of paramount importance. Research to optimise the major process parameters of pepeta processing, such as maturity level and moisture content of immature rice grains, and roasting conditions (i.e., temperature and time), for high nutritional and sensorial quality pepeta product should not be neglected by researchers. Ultimately, assessing proper preservation, packaging, storage conditions and shelf life stability, and possibility for value addition of pepeta product can create new markets in the cities and towns in the country and outside Tanzania. As these food technological problems and solutions are products specific, the social culture, food culture and socio-economic impact of pepeta improvement must be considered by the researchers.
Supporting information S1 Questionnaire. Indigenous processing knowledge and consumers' preferences on pepeta.