Chemical Composition of Major Cassava Varieties in Uganda, Targeted for Industrialisation

Uganda is one of the major cassava producing countries in the world. Currently, utilization of cassava is limited to semi-processed products through the informal sector. Cassava has technological potential as a raw material for agro-industrial products, such as flours for baked products, animal feeds and starch. The aim of this study was to investigate the chemical composition of five major cassava varieties grown in Nebbi distict (Uganda), to assess their potential as industrial raw materials. Analysis of the chemical composition of local (Nyamatia and Nyarukeca) and improved (NASE 3, NASE 14, and NASE 19) cassava varieties was carried out using standard methods. Results showed significant (p < 0.05) differences between the varieties indicating high levels of starch, calcium, magnesium, cyanonenic glucosides and phytates. The cassava varieties contain low levels of protein, lipids and minerals with respect to recommended daily intake of these nutrients. Moisture contents ranged from 5.43 for Nyamatia to 10.87 for NASE 19; ash from 1.05 for Nyamatia to 2.39 for NASE 14; crude fiber from 1.06 for Nyamatia to 1.18 for NASE 19; crude protein from 0.74 for Nyarukeca to 1.51 for NASE 14; crude lipid from 0.39 for Nyamatia to 0.63 for NASE 19; and starch contents from 66.72 for NASE 19 to 84.42 for NASE 3. The mineral contents (mg/kg): calcium ranged from 13.15 for Nyamatia to 16.56 for NASE 3; iron ranged from 0.002 for Nyarukeca to 0.01 for NASE 19; zinc ranged from 0.56 for Nyamatia to 0.87 for NASE 3; magnesium ranged from 3.58 for NASE 19 to 3.88 for Nyarukeca; and copper ranged from 0.002 for Nyamatia to 0.14 for NASE 3. The contents of anti-nutrients (mg/kg): cyanogenic glucosides ranged from 30 in NASE 3 and NASE 19 to 800 in Nyamatia; phytates ranged from 661.33 in Nyarukeca to 984.64 in NASE 3; oxalates ranged from 90.6 in Nyarukeca to 227.8 in NASE 3; and tannin ranged from 0.18 in Nyarukeca to 0.33 in NASE 3. Based on the chemical composition results, all the cassava varieties studied contain higher levels of cyanogenic glucosides than recommended by Ugandan and East African Standards, making them unsafe for direct utilization as food and food raw materials for industries at levels beyond 30% in food formulations. The high starch levels in all the cassava varieties make them valuable raw materials for starch and starch-related industries.


Introduction
Cassava (Manihot esculenta Crantz) is a perennial woody shrub with an edible root, which is widely grown and consumed in tropical and subtropical areas of the world to stave off famine. Uganda is one of the major cassava producers in the world (National Cassava Policy (NCP), 2013). Of the food crops cassava has for a long time been the food security base of the country and has been one of the most important crops in Uganda. However, cassava in Uganda is still a staple food in homes as famine food (security) reserve unlike in Nigeria and Ghana where it is a cash crop and an industrial raw material used in the formulation of various value-added products. Cassava production is in substantial volumes with exceptional production in Northern and Eastern regions of the country. The districts of Lira, Apac, Gulu, Arua, Nebbi, Soroti, Kumi, Tororo, Pallisa, Iganga, and Kamuli are major cassava producing areas (Uganda Census of Agriculture (UCA), 2008/2009; Cassava: Adding Value for Africa (C:AVA), 2007). Currently, cassava is the second most important food crop in Uganda after matooke and its production is estimated at 5.3 million metric tonnes per annum worth about US$5.5 billion (FAOSTAT, 2011). Utilization of cassava is however, limited to semi-processed products through the informal sector and not on industrial scale. The crop has technological potential as a raw material for agro-industrial products, such as flours in baked products, animal feeds and starch (Graffham et al., 2000). Hence, cassava could be exploited to avert food insecurity situation in poor communities and as well contribute to economic development in the country (Sanni et al., 2005).
Cassava is used in many industries including in bakery and confectionaries, brewing, pharmaceuticals, distilleries and packaging industries. Cassava value chain in Nigeria has tremendously contributed to agro-processing industry (Anyiro and Onyemachi, 2014;Ndubueze-Ogaraku and Edema, 2015). Recent developments in Uganda have involved cassava flour in beer brewing at Uganda Breweries Ltd., an innovation beer product known as "engule" that is desired on market nationally, and another trial is underway at Nile Breweries Ltd for another brand of beer. One of the cassava conventional products of economic value for farmers as well as various industries is cassava starch. Variations in starch and sugar concentrations in different cassava varieties would influence to varying levels their organoleptic properties, functionality and physico-chemical properties during their applications in food and industrial processing. Characterization of cassava varieties for their starch and other chemical components would be beneficial in selecting cassava varieties for specific food formulation, processing and ultimately industrial applications. National and regional regulatory authorities have put requirements on the quality of cassava products (Uganda Standards, US 347:2007 for cassava flour; US 597: 2007 for cassava starch). Different industry subsectors have their specific requirements for physical and chemical requirements on cassava flour or cassava starch. Industries in Uganda import starches, dextrin and cassava substitutes from South Africa, Kenya, and Tanzania. Starches largely constitute those from maize and wheat. The importation of starch, dextrin, and cold-setting adhesives leads to loss of large amounts of foreign currency, and to increased unemployment. The extraction of starch from locally grown raw materials will save the country substantial amount of foreign exchange, provide cheap material for research (starch modification), and improve on the knowledge about cassava and its products and eventually create jobs for Ugandans mainly in the starch processing and brewing industries.
Hundreds of cassava varieties are grown in various regions of Uganda. The improved varieties include the NASE family (NASE 1 -NASE 19), NARO CAS 1 and NARO CAS 2; Migyeera, Bukalasa 8, Bukalasa II, and the local varieties include Bao, Okonyoladak, Icilicil, Arawkirra, Tyeno, Oturolak Ayita, Ebwanatereka, Empologoma, Bintiminsi, Serere, Sukari, Kiwoko, , Kulanabwana, Ongada, Akena, Ochide, Angaraba, Malukwa, Nyaraboke, Kakwale, Bamunanika (Ameny, 1990;Nuwamanya et al., 2010). Varietal and environmental factors are known to affect cassava root composition (Corbishley and Miller, 1984). The chemical composition of cassava roots also differs depending on cultural practices like pruning, age and maturity of the root at harvest, storage environment, region, and post-harvest practices. All cassava varieties contain the toxic cyanogenic glucosides Linamarin and Lotaustralin at different concentrations of 150 to 300 mg/kg in peeled root or 300 to 900 mg/kg of the dry matter (Asiedu, 1989;Aalbersberg et al., 1991;Wheatley et al., 1993;Cardoso et al., 2005). Cyanide content of cassava tends to increase during periods of droughts and or prolonged dry weather due to water stress on the plant (Bokanga et al., 1993;Bokanga, et al., 1994). Splittstoesser and Tunya (1992) reported that cassava grown in wet areas contain relatively lower amount of cyanide than those grown in drier areas. Cyanide levels in cassava affect the extent of its acceptability for commercialization. Cyanogenic glucosides may affect local nutrition and may lead to the incidence of cassava related diseases, such as goiter and spastic paraparesis (konzo). However, postharvest handling like fermentation and drying decreases the cyanide amount to acceptable levels. The starch content of cassava largely influences the functional and physico-chemical properties of food systems. These properties include swelling power, swelling volume, solubility, water binding capacity and pasting properties. Afoakwa et al. (2011) suggested that cassava varieties with high moisture and reducing sugar contents would be suitable for fermentation, and could be successfully used as raw material in ethanol, organic acids, lactic bacteria, biofuel industries and may be in latest brewing innovations like the new beer produced at Uganda Breweries Ltd made from cassava flour. Cassava growers, product formulators and developers as well as consumers should therefore know the composition of different cassava varieties. Other processors of non-food products should be able to choose the right varieties for their processes and hence products.
This work characterized cassava varieties based on proximate compositions, concentrations of selected minerals and anti-nutrients in some local and improved high yielding and cassava mosaic disease (CMD) resistant cassava varieties grown in Uganda.

Sample Collection and Sampling
Cassava roots from three improved varieties, NASE 3, NASE 14, and NASE 19, were purchased from a farmer at Pukweru, Panyango sub-county in Nebbi district (Uganda). Roots from two local varieties, Nyamatia and Nyarukeca, were purchased from Gamba village, in Nebbi sub-county, Nebbi district. All the starchy roots were purchased in October, 2016, during the rainy season. The roots were separately washed, weighed and peeled. The peeled roots were sliced using a new grating hand-machine into smaller pieces, dried at 50±2 º C for 24 hours in hot-dry air oven and cooled in air at ambient temperatures. The samples were packaged in dry and new polythene bags purchased from Afroplast Ltd., Kampala, and stored at ambient temperature. Cassava samples were ground to powder using mortar and pestle. The analyses were carried out at Uganda Industrial Research Institute, Chemistry Laboratory and Government Analytical Laboratories.

Determination of Moisture
The method described by AOAC (1995) was adopted where weighed fresh samples were dried in an air oven (Memmet, UFE -600), at 105 º C to a constant weight. The percentage moisture content was calculated as the difference between the fresh and dry weights.

Determination of Ash
The ash content was determined by the drying method described by AOAC (1995) where weighed sample was heated at 550 º C for 5 hrs to ensure proper ashing. The percentage ash content was calculated.

Determination of Crude Fibre
The method described by AOAC (1995) was used for determining the crude fibre of cassava. Two grams of cassava flour was boiled in 100 mL of 0.25 M sulphuric acid solutions under reflux for 30 min. The hot solution was quickly filtered under suction pressure. The residue was thoroughly washed with hot water until it was acid free. The residue was then boiled in 100 mL of hot 0.3 M sodium hydroxide solution under reflux for 30 min and filtered quickly under suction. The insoluble residue was washed with hot water until it was base free. It was dried to a constant weight in an oven at 100º C for 2 hrs, cooled in a desiccator and weighed. The percentage crude fibre content was calculated.

Determination of Nitrogen and Crude Protein
Micro Kjeldahl method (AOAC, 1995) was used. Sample was digested using sulphuric acid and mixed catalyst (96 % CuSO 4 + 3.5 % Na 2 SO 4 0.5 % selenium oxide) in the digestion apparatus (Kjeltec System HT 2, Foss tecator, Hoganä s, Sweden). The distillate, trapped into boric acid solution, was titrated with 0.1 M HCL using a mixture of methyl blue and methyl red as indicators to obtain total nitrogen. Crude protein was calculated using a correction factor of 6.25.

Determination of Crude Lipid
Weighed cassava flour was extracted using petroleum ether in a Sohxlet extraction unit (Soxtec System, Hoganä s, Sweden), according to the method described by AOAC (1995).

Determination of Starch
Total starch content of the samples was estimated using the Anthrone reagent. The sample was first treated with 80:20 (v/v) ethanol/water to extract soluble sugars and the residual starch was hydrolysed with 52 % perchloric acid into monosaccharide (glucose). The glucose was then dehydrated to hydroxymethyl furfural. This compound forms a green coloured product with anthrone. The sugar was colorimetrically determined with 95 % phenol-sulphuric acid by means of a UV-VIS spectrometer (UV -1601, Shimadzu, Japan) at 630 nm. The total starch content was obtained by multiplying the glucose content obtained from the sample using the calibration graph by a factor of 0.9 provided. The calibration graph was prepared by diluting a standard glucose stock solution to obtain a concentration of glucose between 0.02 and 0.1 mg/mL.

Determination of the Minerals
The amounts of Ca, Zn, Fe, Mg, and Cu in cassava was determined using Atomic Absorption Spectrophotometer (AAS) (AAS system, Analyst 400, 2009, Perkin Elmer, Singapore) as described by AOAC (1995). Weighed dry sample was ashed in a muffle furnace at 550 º C for 5 hrs. The ash was dissolved in 5 mL of 20 % HCl. The solution was warmed to dissolve any un-dissolved particles in the residue, filtered through an acid washed filter paper, the filter paper was washed and the solution diluted to volume with potassium chloride solution. The mineral contents: Calcium (Ca), Zinc (Zn), Iron (Fe), Magnesium (Mg), and Copper (Cu) were determined at the respective wavelengths (λ) of 317.0, 213.9, 248.3, 285.2, and 324.8 nm.

Determination of Phytate
The phytate content was determined using the anion-exchange method according to Ma et al. (2005). Cassava flour (1.40 g) was transferred into 100 mL conical flasks. A total of 40 mL of Na 2 SO 4 (100 g/L) and 50 mL of HCl (1.2 %) were added. Flasks were capped and shaken vigorously for 2 hr on a rotator at ambient laboratory temperature. The above mixture was then centrifuged at 500 rpm for 20 min, after which the supernatant solution was filtered through qualitative filter paper No. 4 (Whatman, U.K.). A total of 10 mL of filtered extract was diluted to 30 mL with distilled water after mixing with 1 mL of 0.75 M NaOH and then passed through an anion resin column (resin, AG1-X4, 100-200 mesh, BioRad Laboratory, Inc., CA; column, 0.8 × 10 cm, Beijing Glass Instrumental Factory). The column was washed before use with 20 mL of 0.5 M NaCl solution and deionized water until no Clcan be detected. The absence of Clwas verified by titrating the wash water with neutral silver nitrate using potassium chromate as indicator. Silver chloride is quantitatively precipitated before red silver chromate is formed. After sample application, the column was washed with 15 mL of distilled water and 20 mL of 0.05 M NaCl in order to remove inorganic phosphate. The retained phytic acid (the eluate) from the resin was eluted with 0.7 M NaCl to 25 mL. The post column reagent was made up as a 0.03 % FeCl 3 solution containing 0.3% sulfosalicylic acid. A total of 4 mL of the reagent was added into 5 mL of collected eluate and then centrifuged at 3000 rpm for 10 min. The absorbance of the supernatant was measured at 500 nm using a spectrophotometer (UV-1601, Shimadzu, Japan). A calibration curve for the colorimetric method was obtained by using sodium phytate standards (P-8810 Sigma Co., USA). The Phytate content was calculated using the standard curve. The calibration curve was prepared by taking absorbance of standards with concentration of sodium phytate between 10.0 and 80.0 μg/mL

Determination of Tannin
The tannin content was estimated using Folin -Denis method (Markkar, 1989;Markkar et al., 1993). This is based on the non-stoichiometric oxidation of the molecules containing a phenolic hydroxyl group. Tannin-like compounds reduce phosphotungstomolybdic acid in alkaine solution to produce ahighly coloured blue solution, the intensity of which is proportional to the amount of tannins. Absorbance was taken using spectrophotometer (UV-1601, Shimadzu, Japan) at 760 nm and concentration was estimated from the tannic acid standard curve and results were expressed as milligrams of tannic acid equivalents (TAE) per 100 g of dried sample. The calibration curve was prepared using tannic acid solution (Sigma-Aldrich, B0149-25G, Germany) with concentration range between 25 and 100 μg/L.

Determination of Oxalate
Oxalate was determined by the method of Adeniyi et al. (2009). Two grams of cassava flour was digested with 10 mL 6M HCl for 1 hr and cooled. It was made to the mark in a 250 mL volumetric flask and filtered. 125 mL of the filtrate was measured into beakers and 4 drops of methyl red indicator was added. Concentrated NH 4 OH solution was added drop wise to the test solution until the colour changed from salmon pink to faint yellow and the pH of the solution was determined. Each portion was heated to 90 º C for 4 hr, cooled and then filtered to remove the precipitate. The filtrate was again heated to 90 º C and 10 mL of 5 % CaCl 2 solution was added with continuous stirring. The solution was allowed to stand overnight and then decanted. The precipitate was completely dissolved in 10 mL of 20 % (v/v) hot H 2 SO 4 solution in water. The filtrate was made to 300 mL mark and aliquot of 125 mL of the filtrate was heated until near boiling, which was then titrated against standardized (0.05 M) potassium tetraoxomanganate (VII) to give pink colour (which persists for 30 s) at end point. Oxalate content was calculated as: Oxalate content (g/100g) = T x Vme x Df * 100 (1) ME x MF where : T is titre value of KMnO 4; Vme is volume-mass equivalent (that is, 1 ml of 0.05M KMnO 4 = 0.00228g of anhydrous oxalic acid); Df is dilution factor; Mf is mass of sample; and ME is molar equivalent of KMnO 4 in oxalate concentration (g/dm 3 ).

Determination of Cyanogenic Glycoside
The alkaline pictrate method of Bradbury et al. (1999) was used in the determination of cyanogenic glucoside. Powdered cassava (100.0 mg) was poured on top of a round paper disc containing phosphate buffer at pH 6 and the enzyme linamarase in a flat-bottomed plastic bottle. Clean water (0.5 mL) was added to the sample, and immediately a yellow picrate paper attached to a plastic strip was added. The bottle was immediately closed with a screw capped lid. A blank was prepared the same way except that no sample was added. To verify the method, a standard linamarin paper was used. The bottle was allowed to stand for 24 hrs at ambient temperature. The bottle was then opened and the colour of the picrate paper which developed was compared with the shades of the colour chart. The total cyanide content of the sample was read off from the colour chart.

Determination of total Carbohydrate
The total carbohydrate content was determined by a difference method.

Statistical data analysis
Quantitative data was subjected to analysis using the IBM Statistical Package for Social Sciences (SPSS), version 23. Results are presented as Means ± standard deviations. One way analysis of variance (ANOVA) was used to test for the difference among the varieties. Differences between means were considered significant at p < 0.05.

Proximate Composition
Moisture contents differed significantly (p < 0.05) among the cassava varieties except between NASE 3 and NASE 19 ( Table 1). The moisture contents of the local cassava varieties were lower than those of the improved varieties (5.43 % for Nyamatia and 8.03 % for Nyarukeca as compared to 10.07 % for NASE 3, 8.65 % for NASE 14, and 10.87 % for NASE 19). Sarkiyayi and Agar (2010) reported much lower moisture values of 0.82 % for sweet and 0.14 % for bitter Nigerian cassava varieties. Baah, Oduro and Ellis (2005) reported moisture contents between 6.68 and 10.96% in peeled, washed, grated and de-watered cassava roots. Charles, Sriroth and Huang (2005) reported moisture contents between 9.2 and 12.3 % among cassava varieties grown in Thailand. The differences in the moisture content may be due to their differences in textural structures and constituent solutes. The ash contents differed between the improved and local varieties. The improved cassava varieties had a higher level of ash (2.27% for NASE 3 and NASE 19 and 2.39 % for NASE 14) compared to the local varieties (1.5 % for Nyarukeca and 1.05% for Nyamatia), an indication of a higher mineral content in the improved varieties. Sarkiyayi et al. (2010) reported values of 2.71 % for sweet and 1.85 % for bitter cassava varieties. Safo-kantanka and Acquistucci (1996) reported ash contents between 1.2 and 1.6 % among cassava varieties harvested at 13 months and between 1.2 and 2.1 % among cassava varieties harvested at 6 months. Baah et al., (2005) reported ash contents between 1.15 and 1.2% in cassava varieties. Charles et al. (2005) reported ash contents between 1.3 and 2.8 %. In the present study all the investigated cassava varieties showed low fibre contents (approximately 1 %), which were not significantly different from each other except between NASE 19 and Nyamatia. Sarkiyay et al. (2010) reported fibre contents of 4.40 % for sweet and 4.61% for bitter cassava varieties. Safo-kantanka et al. (1996) reported fibre contents of between 5.4 and 8.3 % among cassava varieties harvested at 13 months; and between 3.9 and 6.1% among varieties harvested at 6 months. Baah et al. (2005) reported fibre contents between 2.9 and 3.64 %. Charles et al. (2005) reported higher fibre contents ranging from 1.5 to 3.5 %.  Sarkiyayi et al. (2010) reported higher protein values (2.69 % for sweet and 3.37 % for bitter varieties). Safo-kantanka et al. (1996) reported lower protein values of between 0.6 % and 1.0 % among six cassava varieties investigated. Baah et al. (2005) reported protein contents of 0.24 and 0.42 %. Charles et al. (2005) reported protein contents between 1.2 and 1.8 %. Nyakaisiki (2016) reported protein values between 0.5 and 1 % on a fresh weight basis in some cassava varieties in western Uganda harvested at 12 months after planting.
Differences in the protein content of cassava varieties may be due to genotype rather than environment. The study shows that cassava is not a major source of protein in the human diet.
Lipid contents were low in both improved (0.48 to 0.63 %) and local (0.39 to 0.48 %) varieties (Table 1). Sarkiyayi et al. (2010) reported very high lipid values (3.92% for sweet and 3.82 % for bitter cassava varieties), while Safo-kantanka et al. (1996) reported lipid values of 1.5 % and 2.2 % for the cassava varieties. Charles et al. (2005) reported lower lipid contents of between 0.1 and 0.8 % in cassava. Difference in lipid content of cassava varieties may be due to the genotype, since these varieties are grown in different countries on the African continent and in Thailand (Asia). In the Ugandan context, cassava is not a good source of lipid in the human diet.
Starch contents varied among the improved (84.42 % for NASE 3,75.25 % for NASE 14,and 66.72 % for NASE 19) and between the local varieties (78.44 % for Nyamatia and 71.75 % for Nyarukeca). Nuwamanya, Baguma, Emmambux, Taylor, and Rubaihayo (2010) reported starch contents ranging between 70.36 and 93.85 % (dry basis) among local and improved cassava varieties grown in Uganda. Safo-kantanka et al. (1996) reported starch contents ranging from 69 to 71 % in cassava roots from Ghana and Nigeria, while Baah et al. (2005) reported starch yields of 68.89 % and 79 % in cassava varieties in Ghana. Nyakaisiki (2016) reported starch contents ranging from 14 to 18 % on fresh weight basis. According to the study, Ugandan cassava varieties are richer in starch than their counterparts in West Africa. Total carbohydrate content was between 83.86 and 91.33, with significant difference between the local variety Nyamatia and all the other varieties both improved and local. Charles et al. (2005) reported a lower range of carbohydrates between 80.1 and 86.3 %. This makes Cassava an important source of carbohydrate in Uganda. Hence, variety but not environment or maturity of the roots may be of significance at the time of harvesting. However, Sarkiyayi et al. (2010) reported total carbohydrate content of 85.46 % in sweet cassava varieties and 86.21% in bitter cassava varieties, showing a significant difference between the two varieties. This may mean that there is a difference in the potential of different varieties to accumulate carbohydrates in their cells.

Minerals
Levels of minerals differed with cassava varieties (Table 2). Zinc, magnesium, copper, and calcium contents were particularly different in the varieties. The mineral contents were generally low except for calcium (13 to 18 mg/100g), and magnesium (3.6 to 3.9 mg/100g). Sarkiyayi et al. (2010) reported higher values for calcium and for iron than levels in this study. Charles et al. (2005) reported much higher levels of calcium (136 to 369 mg/100g) and for magnesium (31 to 43 mg/100g) in cassava varieties grown in Thailand. The copper contents in this study were between 0.002 and 0.14 mg/100g, which are within the limits required by the Ugandan Standards (US 235). Charles et al. (2005) reported higher levels of copper (between 0.037 and 0.057 mg/100g), zinc (between 13 and 19 mg/100g), and iron (between 29 and 40 mg/100g). Thai cassava varieties are thus richer in minerals than Ugandan varieties. Cassava roots may be a reasonable source of calcium and magnesium in the diet of Ugandans compared to other minerals evaluated.

Anti-Nutritional Compounds
Levels of the cyanogenic glycosides were significantly higher in the local varieties compared to the improved varieties (Table 3). Sarkiyayi et al. (2010) reported much lower cyanogenic glucoside values of 4.6 mg/kg and 6.5 mg/kg for sweet and bitter cassava varieties respectively. Nyakaisiki (2016) reported cyanogenic glucoside contents ranging between 28 and 53 mg/kg on fresh weight basis. Charles et al. (2005) reported lower values for cyanide potential in cassava varieties from Thailand, ranging from 8.33 to 28.8 mg/kg on dry weight basis. The cyanide potential ranged between 26.9 and 28.8 mg/kg in bitter cassava varieties and between 8.33 and 12.5 mg/kg in sweet cassava varieties. According to Uganda Standards, the total hydrocyanic acid content of cassava flour shall not exceed 10 mg/kg (US 347: 2007). The high data for cyanide obtained in this study may be due to the effect of interfering compounds (sulphite, sulphide, iodine, chloride and thiosulphate). The local varieties are also bitter, implying that the level of cyanogenic glucosides in a cassava root is directly related to bitterness. Since cyanogenic glucosides release the toxic cyanide as a breakdown product, bitterness of a cassava root is directly related to its toxicity.
Phytate contents were high for all cassava varieties and differed significantly. It was higher in improved varieties compared to local varieties (Table 3). Sarkiyayi et al. (2010) reported much higher phytate values of 2,160 mg/kg and 3, 040 mg/kg for sweet and bitter cassava varieties respectively. Charles et al. (2005) reported phytate levels ranging between 950 and 1, 360 mg/kg in cassava with no difference between sweet and bitter varieties. Phytate may impair the bioavailability of iron, calcium, magnesium, and zinc in the diets of people dependent on cassava as a staple food. However, phytate may play the role of an antioxidant by sequestering iron and thus hinder the formation of free radicals.
Oxalate contents differed significantly among the cassava varieties in this study. The levels were generally higher in improved varieties compared to local varieties (Table 3). Sarkiyayi et al. (2010) reported oxalate contents of 220 mg/kg in sweet and 440 mg/kg in bitter cassava varieties. Oxalates may chelate minerals such as calcium, zinc and iron and therefore prevent their absorption and utilization by the human body.
Tannin contents differed among the cassava varieties studied except among NASE 14, NASE 19, and Nyamatia (Table 3). The tannin levels were generally higher in improved cassava varieties than in the local varieties. The levels in this study were lower than those reported by Sarkiyayi et al. (2010). In cassava, wound responses may lead to the formation of condensed tannins in the roots which cause a discolouration of the vascular tissue and storage parenchyma. The levels of the anti-nutrients in this study (cyanide, phytate, oxalate and tannin) may be significantly reduced by processing of cassava roots, such as cooking, fermentation and soaking, and hence render the processed roots safe for human consumption.

Conclusions
The proximate composition, mineral composition, and contents of anti-nutritional factors in cassava differ between improved and local varieties. The improved varieties are generally sweet while the local varieties are bitter. The protein contents are within the range required by some industries utilizing cassava flour in their processes. The cyanogenic contents of all the cassava varieties studied are above the value stipulated in the Uganda Standards for cassava flour making them unsuitable for use in their primary form. The generally low mineral contents will necessitate the fortification of the flour used for production of human food. The high starch contents of the cassava varieties in this study make the cassava roots valuable raw materials for industrial utilization.