Physicochemical, biochemical, and functional characterization of improved cassava varieties resistant to cassava brown streak disease

1 Introduction

Cassava (Manihot esculenta Crantz) is a semi-perennial starchy root crop that supports food security, income generation, and agricultural resilience across tropical regions. Africa accounts for most global cassava production, and the Democratic Republic of the Congo (DRC) is among the continent’s leading producers (Chikoti and Tembo, 2022; Borku et al., 2025). Cassava is an essential staple because it tolerates variable rainfall, low-input conditions, and relatively poor soils, thereby remaining productive in marginal environments (Tabaglio et al., 2023; Borku et al., 2025; Mondo et al., 2026). Nutritionally, cassava is primarily an energy food rich in carbohydrates, while protein and lipid concentrations are generally low (Nilusha et al., 2021; Borku et al., 2025).

Despite its importance, cassava production is severely constrained by cassava mosaic disease (CMD) and cassava brown streak disease (CBSD). Across Africa, CMD and CBSD may reduce cassava harvests by approximately 30–40% and up to 70%, respectively, under severe disease pressure (Chikoti and Tembo, 2022). In eastern DRC, CBSD has expanded into previously unaffected territories, threatening the marketability and nutritional value of storage roots (Casinga et al., 2021). Disease-resistant or tolerant varieties are, therefore, central to disease management, but their promotion must also consider agronomic and postharvest performance. Field research in the highlands of South-Kivu confirms that the use of improved varieties can increase cassava productivity, especially when combined with appropriate soil-fertility management (Munyahali et al., 2023).

Nevertheless, farmer adoption of improved cassava varieties remains limited when new genotypes do not match local expectations for processing, end-user characteristics, product quality, and market demands. Evidence from eastern DRC shows that adoption and varietal preference depend on both production advantages and farmer-valued traits (Mondo et al., 2019). Processor and consumer preferences are also strongly influenced by flour yield, functional, and sensory qualities (Chimphepo et al., 2021; Chijioke et al., 2021; Awoyale et al., 2023). Consequently, comparative postharvest characterisation is needed before varieties are widely recommended.

This study contributes to the ongoing efforts by the International Institute of Tropical Agriculture (IITA) to develop and promote CBSD-resistant cassava varieties in eastern DRC in order to sustain production amid outbreaks of viral diseases that threaten the food and income security of smallholder cassava farmers. Therefore, this study compared four improved cassava varieties with two locally preferred varieties under South-Kivu cultivation conditions. It characterized physicochemical and nutritional traits, assessed functional and temperature-dependent processing properties, and examined multivariate relationships among traits to identify CBSD-resistant varieties with the best balance of food safety, nutritional value, and technological suitability relative to locally preferred checks.

2 Materials and methods

2.1 Plant material

Six cassava varieties were evaluated: four improved CBSD-resistant varieties and two popular local varieties collected directly from farmers in Kavumu, eastern DRC. Based on local breeder documentation (SENASEM, 2026) and farmers’ classification at sampling, Mundola and Mulezi are bitter, whereas Sikirou and Bukenge are sweet. Among the local varieties, Nambiyombiyo (also known as Mwambiyo-mbiyo) is bitter, whereas Ngorodima is sweet. The two local varieties are preferred by producers for their desirable sensory attributes. Additional varietal characteristics are presented in Table 1.

Table 1. Characteristics of cassava varieties used in the study.

Source: SENASEM (2026); authors’ field records.

2.2 Preparation of cassava roots for laboratory analyses

The improved varieties were harvested from cassava multiplication fields at INERA-Mulungu, South-Kivu, eastern DRC, where they were being multiplied for subsequent distribution to cassava seed multipliers. On the other hand, local cassava samples were collected from non-preferentially selected plants in farmers’ fields near INERA-Mulungu to minimize agroecological differences between improved and local varieties. As the crops were not grown specifically for this study, no controlled field experimental design was imposed. After harvest, roots were randomly sampled to make a total of 60 kg from each variety. Upon arrival at the laboratory, three fresh roots were washed with potable water to remove adhering soil and debris, and then peeled using clean stainless steel knives. The roots were cut into small, uniform pieces to facilitate handling and ensure an even distribution of constituents and then placed in labelled, airtight food-grade plastic bags for direct determination of pH, titratable acidity, moisture content, and hydrogen cyanide (HCN).

For the remaining physicochemical and functional analyses, sound fresh roots were randomly selected, peeled, and grated into thin lamellae to increase the surface area available for moisture removal, and dried in an electric cabinet dryer (ST-01 T, Ultron Technology, Guangdong, China) at 60 °C for 6 h (FAO, 1994). After complete drying, the samples were finely milled using a FOUFOU-MAIS hammer mill (D03 2FM10T, Deklerck-Bexen, Brussels, Belgium) fitted with an appropriate screen to obtain homogeneous flour. The milled samples were then stored in airtight plastic packaging protected from light, air, and moisture until analysis. This procedure yielded stable samples for analysis, as illustrated in Figure 1 (Nainggolan et al., 2024).

Moisture content, dry matter, pH, total titratable acidity (TTA), hydrogen cyanide (HCN), and ash were determined using three biological replicates per variety, each biological replicate corresponding to a fresh root from a different plant. Protein, crude fiber, lipid, starch, amylose, and sugar determinations were likewise performed on separate processed samples from these three plants. In contrast, functional properties were determined on composite flours per variety, with each functional assay performed in technical triplicate.

2.3 Physicochemical analyses

Moisture content was determined gravimetrically following the Association of Official Agricultural Chemists (AOAC) International (2019). Approximately 5 g of the sample was weighed into a previously tared dish and dried in an oven (UN30plus, Memmert GmbH + Co. KG, Schwabach, Germany) at 105 °C for 4 h. The samples were cooled in a desiccator for 30 min and reweighed. Moisture content was calculated using Equation 1.

Moistur e ( % ) = P 1 − P 2 P 1 × 100 (1)

Where: P1 is the sample mass before drying, and P2 is the sample mass after drying.

The pH was measured using a previously calibrated pH meter (LUTRON PH-230SD, Lutron, Taipei, Taiwan). A 10 g sample was homogenised in 100 mL of distilled water with agitation for 30 min and then filtered. The pH of the filtrate was measured directly (Babatuyi et al., 2024).

HCN content was determined on a fresh-weight basis using a method adapted from ISO 2164:1975. A 10 g sample was hydrolysed in 200 mL of water for 1 h and then distilled. The distillate was collected in 2% boric acid solution and titrated with silver nitrate in an ammoniacal medium in the presence of potassium iodide. HCN content was calculated as follows (Equation 2):

HCN ( mg kg ) = 0.54 ( V 0 − V 1 ) × 250 100 × 1000 m = 1350 ( V 0 − V 1 ) m (2)

Where: V₁ is the volume of silver nitrate solution used for the blank, and V₀ is the volume used for sample titration.

Titratable acidity was determined in accordance with ISO 750:1998. An aqueous extract (10 g/100 mL) was prepared, filtered and titrated with 0.1 N NaOH using phenolphthalein as the indicator(Equation 3):

TA = V NaOH × N NaOH × EW Sample volume ( mL ) × 10 (3)

Where: V_NaOH is the volume of sodium hydroxide (NaOH) titrant used (mL); N_NaOH is the normality of the NaOH solution (e.g., N = 0.1 for a 0.1 M solution); EW is the equivalent weight of the acid (90 for lactic acid in the present calculation); Sample volume is the volume of the analysed extract (mL); and 10 is the unit-conversion factor.

Protein content was determined by formol titration as described in standard food-analysis procedures (Nielsen, 2017). Following alkaline extraction, amino groups reacted with formaldehyde and were titrated with 0.1 N NaOH (Equations 4,5):

% N = ΔV × N NaOH × 14.007 m (4)

% Protein = % N × F (5)

Where: ΔV is the difference between the NaOH volume consumed by the sample and the blank (mL); N_NaOH is the normality of NaOH (eq/L), and 14.007 is the atomic mass of nitrogen (mg/meq); m is the mass of the analysed sample (g); and F is the nitrogen-to-protein conversion factor (6.25 for total protein and 5.7 for starchy matrices such as cassava).

Total ash content was determined by incineration at 550 °C for 6 h following Association of Official Agricultural Chemists (AOAC) International (2019), using the following (Equation 6):

Ash ( % ) = R e sidual mass Initial mass × 100 (6)

Crude fibre was determined using the Weende method (AOAC 962.09). The sample underwent sequential acid and alkaline digestion followed by incineration [Association of Official Agricultural Chemists (AOAC) International, 2019] (Equation 7):

Crude fibre ( % ) = P 1 − P 2 P × 100 (7)

Where: P is the initial dry sample mass (g), P₁ is the mass of the dried residue after sequential acid and alkaline digestion (g), and P₂ is the mass of ash remaining after incineration (g).

Total lipids were extracted with petroleum ether and quantified gravimetrically according to AOAC 920.39 [Association of Official Agricultural Chemists (AOAC) International, 2019] (Equation 8):Where: m_{fat residue} is the mass of petroleum-ether-extractable lipid residue (g), and m_sample is the initial dry sample mass used for extraction (g).

Lipids ( % ) = m fat residue m sample × 100 (8)

Total carbohydrates were determined using the enzymatic glucose oxidase-peroxidase (GOD-POD) method, with absorbance measured at 620 nm. Reducing and total sugars were quantified using the Luff-Schoorl procedure described in Association of Official Agricultural Chemists (AOAC) International (2019), as follows (Equation 9):

Reducing sugars ( % ) = ( V 0 − V ) × N × 0.00342 × 100 m (9)

Where: V₀ is the volume of thiosulfate used for the blank (mL); V is the volume of thiosulfate used for the sample (mL); N is the normality of the thiosulfate solution (0.1 N); 0.00342 is the mass of glucose (g) equivalent to 1 mL of 0.1 N Na₂S₂O₃; and m is the mass of the analysed sample (g).

After acid hydrolysis, total sugars were determined using the same method (Equation 10):

Total sugars ( % ) = ( V 0 − V t ) × N × 0.00342 × 100 m (10)

Where: V_t is the volume of thiosulfate consumed after acid hydrolysis.

Starch content was calculated after acid hydrolysis as follows (Equation 11):The factor 0.90 corrects for the difference in molar mass between starch and glucose. The procedure followed the relevant AOAC methods for reducing sugars and starch after acid hydrolysis [Association of Official Agricultural Chemists (AOAC) International, 2019].

Starch ( % ) = % Glucose × 0.90 (11)

Amylose was determined iodometrically following AOAC 996.11 [Association of Official Agricultural Chemists (AOAC) International, 2019]. Bound iodine (I_b) was titrated with thiosulfate (Equation 12):

I b = ( V b − V e ) × N tthio × 1.269 mg (12)

Where: V_b is the thiosulfate volume used for the blank (mL), V_e is the thiosulfate volume used for the sample extract (mL), N_thio is the normality of the sodium thiosulfate solution, and 1.269 is the iodine-equivalence factor used in the calculation.

The result was converted to amylose mass using the calibration curve (Equation 13):Amylose content was then expressed as a percentage (Equation 14):In Equations 13, 14, I_b is the bound iodine calculated from Equation 12; k is the calibration coefficient obtained from the amylose standard curve; m_amylose is the amylose mass derived from the calibration curve (mg); and m_starch is the starch mass represented in the analysed aliquot (mg).

Amylose ( mg ) = I b k (13)

Amylose ( % ) = m amylose m starch × 100 (14)

2.4 Functional properties

Water absorption capacity (WAC) was determined using a standard centrifugation procedure reported for cassava flour by Nilusha et al. (2021). One gram of the sample was mixed with 10 mL of distilled water in a centrifuge tube and shaken for 30 s. After standing for 30 min at room temperature, the mixture was centrifuged at 3,000 rpm for 15 min. The supernatant was discarded, and the resulting gel was weighed. WAC was then calculated using the following (Equation 15):or, equivalently (Equation 16):

WAC = Mass of gel after decantation − Dry sample mass Dry sample mass (15)

WAC = Mass ofwater absorbed ( g ) Samp l e mass ( g ) (16)

Equation 15 was used for the calculations in this study. The ratio was multiplied by 100 to express water absorption capacity as a percentage; Equation 16 represents the same relationship before percentage conversion. In these equations, m_gel is the mass of the hydrated gel after decantation (g), m_dry is the initial dry sample mass (g), and m_water absorbed = m_gel − m_dry.

Least gelation concentration (LGC) was determined using the procedure reported for cassava flour by Nilusha et al. (2021). Aqueous flour suspensions ranging from 1 to 20% (w/v) were prepared in 10 mL of distilled water. After homogenisation, the samples were heated in a boiling water bath for 1 h, cooled to room temperature, and held at 4 °C for 2 h. The lowest concentration that formed a stable gel when the tube was inclined at 45° was recorded as the LGC. Here, LGC refers only to the minimum flour concentration needed to form a stable gel under the specified assay conditions and does not independently indicate overall gel quality.

Bulk density was determined by gently introducing the powder into a 10 mL graduated cylinder without compaction and recording its mass (Nilusha et al., 2021) (Equation 17):

Bulk density = Flour mass ( g ) Occupied volume ( mL ) (17)

Swelling power and solubility index were determined using the procedure described by Nilusha et al. (2021). One gram of sample was suspended in 10 mL of distilled water and heated at 60, 80, 90 or 100 °C for 30 min. After cooling, the suspension was centrifuged at 3,000 rpm for 15 min. Swelling power was calculated as follows (Equation 18):

Swelling power ( g g ) = m wet gel m flour (18)

Where: m_wet gel is the mass of the gel after centrifugation, and m_flour is the initial flour mass (1 g).

The supernatant was collected, oven-dried, and weighed to determine the soluble fraction. The solubility index was calculated as follows (Equation 19):

Solubility index ( % ) = m dissolved solids m initial flour × 100 (19)

This procedure enabled comparison of starch swelling and solubilisation as a function of temperature, two key indicators of functional performance.

Particle-size distribution was determined by dry sieving through a series of standard sieves (600, 250, 150 and 45 μm) (Akonor et al., 2023). A 100 g sample was sieved for 10 min, and the fraction retained on each sieve was weighed (Equation 20):

Retained fraction ( % ) = m i m 0 × 100 (20)

Where: m_i is the mass retained on sieve i and m₀ is the initial sieved mass (100 g).

Dispersibility was evaluated by mixing 5 g of the sample with 100 mL of distilled water and agitating for 2 min. After standing for 1 h, the sediment volume was recorded (Nilusha et al., 2021) (Equation 21):

Dispersibility ( % ) = ( 1 − V s 100 ) × 100 (21)

Where: Vs is the sediment volume (mL). For example, a sediment volume of 30 mL corresponds to 70% dispersibility, indicating that most of the flour remains well dispersed.

Apparent viscosity was measured using a digital rotary viscometer (NDJ-5S, VEVOR, China). A suspension containing 3 g of sample on a dry basis in 25 mL of distilled water was evaluated at 50, 65, 80, and 95 °C (Akonor et al., 2023). The measured parameters included peak viscosity, pasting temperature, breakdown viscosity, setback viscosity, and final viscosity.

2.5 Data analysis

Differences among the six cassava varieties were assessed using one-way analysis of variance (ANOVA), with variety as the factor, at a 5% significance level (p < 0.05). Normality and homogeneity of variance were assessed before analysis, and Fisher’s least significant difference (LSD) post hoc test was used for post-hoc comparisons when significant varietal effects were detected. For compositional traits, each variety was represented by three biological replicates, each derived from a fresh root collected from a different plant; protein, crude fibre, lipid, starch, amylose, and sugar contents were determined using processed samples from these same plants. Functional properties were measured in technical triplicate using composite flour samples. Pearson’s correlation coefficients were calculated using 18 observations (three biological replicates per variety across six varieties) to assess linear relationships among quantitative physicochemical, biochemical, and functional variables. Principal component analysis (PCA) was also performed on the six variety means to reduce data dimensionality, visualise multivariate relationships, and identify associations among varieties and measured traits. Statistical analyses were performed using R (R Core Team, 2020).

3 Results and discussion

3.1 Physicochemical characteristics of cassava roots

Physicochemical properties of roots from different cassava varieties are presented in Table 2. There were highly significant differences (p < 0.001) among varieties for all measured physicochemical parameters, confirming a strong varietal effect on the cassava roots’ physicochemical properties. The pH was generally low for the improved varieties compared to the local checks; it ranged from 6.63 in Bukenge to 7.36 in Ngorodima. The improved varieties Sikirou (6.78), Mundola (6.92), and Mulezi (7.21) had intermediate values, some of which were comparable with those of the local varieties Nambiyombiyo (7.31) and Ngorodima (7.36). The observed pH variation indicates biochemical differentiation among genotypes and may reflect differences in organic-acid composition.

Table 2. Physicochemical properties of roots from different cassava varieties.

^***, very highly significant (p < 0.001); CV, coefficient of variation. Within a column, means followed by the same letter are not significantly different according to the least significant difference (LSD) test at 5% p-value level. ^§The moisture content and HCN values are reported on a fresh-weight basis.

Moisture content differed significantly (p < 0.001) among varieties, ranging from 65.80% in Mulezi to 71.70% in Ngorodima. Improved varieties Mulezi (65.80%) and Mundola (66.50%) contained less water than the local varieties Nambiyombiyo (70.53%) and Ngorodima (71.70%). This attribute is of interest in processing as firmer, less watery roots are generally associated with higher dry-matter recovery and greater yields of processed products (Akonor et al., 2023). Nilusha et al. (2021) also identified moisture and dry matter as key determinants of flour stability and technological performance.

Total ash was highest in the local variety Nambiyombiyo (10.22%), followed by Mundola (9.02%) and Ngorodima (8.86%). In contrast, Bukenge (5.38%) and Sikirou (5.31%) had the lowest values, indicating marked genotypic variations in overall concentration of total mineral elements and non-volatile inorganic matter (Ayetigbo et al., 2018). For instance, such high ash content among local varieties reflects their genotype-specific capacities for mineral uptake from the soil where they were grown.

Titratable acidity ranged from 0.13 g/L in Sikirou to 1.03 g/L in Mundola. Local varieties had intermediate values of 0.77 g/L for Nambiyombiyo and 0.53 g/L for Ngorodima. These differences demonstrate a strong genotypic influence on root acidity and biochemical composition. They may also contribute to differences in taste, postharvest stability, and suitability for fermentation. Physicochemical traits such as acidity should, therefore, be incorporated into technological assessments because processor and consumer preferences are strongly shaped by texture, flour yield, and sensory characteristics of cassava products (Baguma et al., 2022; Chimphepo et al., 2021; Awoyale et al., 2023).

Hydrogen cyanide content varied markedly, from 36.94 mg/kg in Sikirou to 128.84 mg/kg in Nambiyombiyo. In particular, the improved varieties Sikirou (36.94 mg/kg) and Mulezi (53.79 mg/kg) had substantially lower HCN concentrations than local varieties. This finding highlights a potential food-safety advantage of improved varieties because cyanogenic compounds are major antinutritional and toxicological concerns in cassava (Quinn et al., 2022; Baguma et al., 2022; Bamidele, 2025). The relatively low HCN contents of Sikirou and Mulezi may facilitate processing and reduce consumer risks. By contrast, the high HCN levels in Nambiyombiyo, Mundola, and Bukenge require effective detoxification through a combination of different processing steps such as soaking, grating, dewatering, fermentation, drying, and adequate cooking before consumption.

3.2 Biochemical characteristics of cassava roots

All biochemical traits differed significantly among varieties (p < 0.001), demonstrating a strong genotypic influence on cassava root composition (Table 3). Crude fibre ranged from 1.95% in Sikirou to 2.54% in Ngorodima. Improved varieties Bukenge (2.53%) and Mulezi (2.53%) had fibre contents comparable with those of the local checks. Although fibre levels were generally low to moderate, the observed differences may influence texture, digestibility, milling behaviour, and flour performance (Nilusha et al., 2021). The comparatively high values in Ngorodima, Bukenge, and Mulezi may indicate a more fibrous cellular structure, with potential consequences for particle-size distribution and the acceptability of derived foods.

Table 3. Biochemical properties of test cassava varieties.

^***, very highly significant (p < 0.001); CV, coefficient of variation. Within a column, means followed by the same letter are not significantly different according to the least significant difference (LSD) test at 5% p-value threshold. Protein, crude fiber, lipids, carbohydrates, sugars, starch, and amylose were determined on grated, dried, and milled samples.

Protein contents were low, as expected for cassava, but differed significantly among genotypes, ranging from 1.13% in Ngorodima to 2.07% in Bukenge. Bukenge, an improved variety, had the highest protein content, exceeding those of Nambiyombiyo (1.73%) and Mulezi (1.23%). These results confirm that cassava is primarily an energy source rather than a major protein source (Nilusha et al., 2021; Borku et al., 2025).

Carbohydrates were the predominant fraction, ranging from 79.97% in Bukenge to 87.90% in Mulezi. Improved varieties Mulezi (87.90%) and Sikirou (85.00%) had concentrations comparable with or higher than those of the local varieties Nambiyombiyo (81.97%) and Ngorodima (85.97%). This predominance confirms cassava’s primary role as an energy food (Nilusha et al., 2021; Borku et al., 2025). Reducing sugars ranged from 4.20% in Bukenge to 8.97% in Nambiyombiyo, while total sugars reached 12.50% in Nambiyombiyo and 11.50% in Ngorodima. Improved varieties had moderate sugar concentrations, including 5.05% in Mulezi and 5.50% in Sikirou. Thus, local varieties were richer in soluble sugars than improved varieties. This may partly explain their sensory preference where a mildly sweet taste is valued. However, high sugar concentrations can also influence fermentation, non-enzymatic browning, and the postharvest stability of processed products.

Starch content differed significantly among varieties, ranging from 27.90% in Bukenge to 33.30% in Mulezi. Mulezi (33.30%) and Sikirou (31.50%) had values comparable with or higher than those of the local checks Nambiyombiyo (29.70%) and Ngorodima (30.60%), confirming the strong technological potential of improved varieties. High-starch roots generally produce more cohesive pastes and a mealier texture. This agrees with Nilusha et al. (2021), who identified starch content as a major determinant of cassava processing behaviour. Widowati et al. (2025) further associated starch characteristics with flour gelatinisation and functional performance.

Amylose content ranged from 25.10% in Bukenge to 37.80% in Nambiyombiyo. Mundola (37.50%) had a high concentration comparable with that of the best-performing local variety. This variation is technologically important because amylose affects gelatinisation, retrogradation, gel firmness, and texture of cassava-based foods (Akonor et al., 2023; Widowati et al., 2025). The high amylose contents of Nambiyombiyo and Mundola may promote firmer gels, whereas the lower contents of Bukenge, Mulezi, and Ngorodima may produce softer or less cohesive textures, depending on the processing method.

Lipid contents varied only slightly, from 0.42 to 0.44%, although the varietal effect showed highly significant differences (p < 0.001). Improved varieties were relatively uniform (0.43–0.44%), whereas Nambiyombiyo had the lowest value (0.42%). These low concentrations confirm that cassava is naturally low in fat (Nilusha et al., 2021; Borku et al., 2025). Although lipids make only a minor contribution to cassava’s energy value, their low concentration may favour flour stability by limiting oxidative deterioration.

Total sugar content differed significantly among varieties (p < 0.001), ranging from 8.10% in Mundola to 12.50% in Nambiyombiyo. Local varieties contained 11.50–12.50%, whereas improved varieties contained 8.10–9.50%. This pattern suggests that local varieties may be favoured for fresh market with sweetness and fermentation potential, whereas improved varieties are distinguished more by starch content, processing yield and, in some cases, lower HCN. Cassava-variety evaluation should, therefore, extend beyond agronomic yield to include biochemical, nutritional, and technological attributes relevant to farmers, processors, and consumers.

3.3 Functional properties of cassava flours

Significant varietal differences were observed in water absorption capacity (WAC) (p < 0.001), least gelation concentration (LGC) (p = 0.0133), and bulk density (p < 0.001), whereas dispersibility did not differ significantly among varieties (Table 4). These results indicate that genotype substantially influenced the functional behaviour of cassava flour, although the magnitude of the varietal effect depended on the functional attribute considered. WAC ranged from 106.67 to 210.00%, with Nambiyombiyo (210.00%) and Sikirou (208.00%) showing the highest values whereas Ngorodima had the lowest WAC (106.67%). These values were generally higher than those reported by Awoyale et al. (2015) for three biofortified cassava root starches, for which WAC ranged from 75.86 to 86.50%. Their study also reported dispersibility of 80.00–86.00%, swelling power of 6.27–9.59%, solubility index of 1.30–1.90%, and LGC of 4.01–4.06%. The higher WAC observed in the present study may reflect differences between whole cassava flour and isolated starch, as well as differences in genotype, particle characteristics, starch damage, and processing conditions. WAC is technologically important because it influences hydration requirements, dough consistency, gel formation, and product texture. Thus, the high WAC of Sikirou, Mundola, and Mulezi may be advantageous in products requiring substantial water retention, although a high WAC should not be considered universally superior because its desirability depends on the intended product.

Table 4. Functional properties of flours from different cassava varieties.

ns, not significant (p > 0.05); , significant (p* < 0.05); **, very highly significant (p* < 0.001); CV, coefficient of variation. Within a column, means followed by the same letter are not significantly different according to the least significant difference (LSD) test at 5% p-value threshold.

Least gelation concentration (LGC) differed significantly among varieties. Mundola and Nambiyombiyo exhibited the lowest LGC (2.00 ± 1.00%), followed by Sikirou (5.00 ± 1.00%), while Bukenge, Mulezi, and Ngorodima required 6.00 ± 2.00% to form a stable gel. The 2.00% value obtained for Mundola and Nambiyombiyo is comparable to the low values reported for cassava starch by Akinwale et al. (2017), whose LGC ranged from 2.00 to 2.67%. Their results, therefore, provide strong comparative evidence for the good gel-forming capacity observed in these two varieties. By contrast, Awoyale et al. (2015) reported LGC values of 4.01–4.06% for biofortified cassava starches, while Lawal et al. (2021) reported values of ~4.01–4.06% in cassava materials used for pasta development. Even higher values (14–16%) were reported by Awoyale et al. (2017) for commercial dried cassava products, particularly toasted products. The substantially lower LGC in the present study may reflect differences in processing history and the absence of the intense thermal transformations associated with some commercial products. The low LGC of Mundola is particularly interesting because this improved CBSD-resistant variety combined efficient gel formation with a relatively high amylose content (37.50%). Nambiyombiyo showed a similar profile, with 37.80% amylose and an LGC of 2.00%. In contrast, Bukenge and Mulezi had approximately 25.10% amylose and LGC values of 6.00%, while Sikirou had 32.10% amylose and an intermediate LGC of 5.00%. The strong negative correlation between amylose and LGC (r = −0.809) supports an association between starch composition and gelation behaviour. However, this relationship should not be interpreted as causal evidence that amylose alone determines LGC, since starch damage, amylopectin structure, particle size, and non-starch components may also influence gel formation. This exploratory pattern is consistent with a possible role of amylose in gel formation, but confirmation requires a larger set of genotypes and direct rheological measurements. Rasaq et al. (2020) similarly demonstrated that cassava variety and production factors influence the functional properties of high-quality cassava flour.

Bulk density varied from 0.38 to 0.55 g/mL, with Nambiyombiyo showing the highest value (0.55 g/mL) and Mulezi the lowest (0.38 g/mL). These values are comparable to the loose bulk densities of 0.51–0.53 g/mL reported by Lawal et al. (2021) for cassava flours, although direct comparison should consider differences in measurement conditions. Awoyale et al. (2017) reported bulk-density values corresponding to 0.47–0.58 g/mL among commercial dried cassava products, placing most of the test varieties within a similar practical range. Bulk density is relevant to packaging, transportation, storage, and ingredient dosing because denser flour occupies less volume for a given mass. The relatively high density of Nambiyombiyo may, therefore, favour compact packaging, whereas the lower density of Mulezi indicates a more voluminous flour.

Dispersibility ranged from 85.70 to 88.67% and did not differ significantly among varieties. Bukenge had the highest value (88.67%), whereas Nambiyombiyo had the lowest (85.70%). These values are slightly higher than the 80.00–86.00% range reported by Awoyale et al. (2015) for biofortified cassava starches, suggesting generally favourable reconstitution behaviour in the present materials. The absence of a significant varietal effect indicates that dispersibility was comparatively stable despite differences in WAC, LGC, and bulk density. Such relatively high dispersibility may be beneficial in products requiring rapid and uniform reconstitution, including porridges and other flour-based preparations.

Temperature-dependent behaviour further demonstrated significant varietal differences in swelling power (SP) and solubility index (SI) (p < 0.001), while apparent viscosity differed significantly only at 50 °C (p = 0.0059). Ngorodima showed the highest swelling at 60–80 °C, reaching approximately 9.90–10.30 g/g, whereas Bukenge showed lower values of approximately 3.10–4.10 g/g. At 100 °C, however, Mundola and Mulezi exhibited the highest swelling values (7.50 and 7.13 g/g, respectively). This temperature-dependent reversal indicates differences in starch hydration, granule stability and disruption. The SP values observed are broadly comparable with the 6.27–9.59% range reported by Awoyale et al. (2015), although differences in analytical expression and sample type should be considered when making direct numerical comparisons. Solubility also increased with temperature, with marked varietal differences (p < 0.001). At 100 °C, Sikirou, Mundola, and Bukenge showed comparatively high solubility, indicating greater release of soluble starch fractions during heating. This behaviour is technologically relevant because starch swelling and solubilisation contribute to paste formation and ultimately influence texture. The present correlations between SI100 and dispersibility (r = 0.627), reducing sugars (r = −0.881), total sugars (r = −0.765), and pH (r = −0.787) further indicate that thermal solubility was associated with several compositional characteristics, although these relationships should be interpreted as associations rather than direct causal effects. Apparent viscosity was highest at 50 °C for Sikirou (10.83 Pa·s), Bukenge (10.60 Pa·s), and Ngorodima (10.40 Pa·s), but varietal differences disappeared at 65–95 °C. This convergence suggests that prolonged heating reduced differences in hydration and starch structural response. Such behaviour is consistent with the known influence of starch composition, granule structure, and swelling on cassava pasting properties.

Finally, particle-size distribution differed significantly among varieties (p < 0.001). Ngorodima produced a relatively fine flour, with 56.20% of particles in the 45–150 μm fraction and 27.15% below 45 μm, whereas Bukenge contained greater proportions of coarse particles (>600 μm: 15.76%; 250–600 μm: 26.10%). The finer particle distribution of Ngorodima may facilitate water penetration and partly explain its rapid swelling at intermediate temperatures, whereas the coarser structure of Bukenge may contribute to slower hydration. Overall, no variety was universally superior across all functional properties. Mundola combined high WAC, low LGC and high amylose; Sikirou showed high WAC and favourable thermal behaviour; Mulezi combined high WAC with high starch content and low bulk density; Nambiyombiyo combined high WAC, low LGC, and high bulk density; and Ngorodima showed rapid swelling and a fine particle-size distribution. These complementary profiles indicate that variety selection should be based on the intended technological application, rather than on a single functional parameter. Importantly, the results demonstrate that improved CBSD-resistant varieties, particularly Mundola, can retain functional characteristics comparable to or better than those reported for several cassava materials in previous studies.

Figure 2 presents swelling power, solubility index, viscosity, and particle-size distribution for the different cassava varieties. There were significant varietal differences (p < 0.001) in swelling power (SP), solubility index (SI), and particle-size distribution. Viscosity differed significantly only at 50 °C and not at higher temperatures. These results confirm that cassava-flour functionality is strongly affected by genotype and processing temperature. Swelling power and solubility index are central indicators of starch behaviour in heated aqueous systems because they reflect the capacity of starch granules to absorb water, expand, release soluble constituents, and develop viscosity (Nilusha et al., 2021). The observed differences, therefore, indicate varietal variation in starch structure, amylose content, particle size, and the organisation of flour components.

Swelling power varied markedly with both temperature and variety. At intermediate temperatures (60–90 °C), the local variety Ngorodima consistently had the highest values (9.90–10.30 g/g at 60–80 °C; p = 0.0029), indicating rapid early hydration. In contrast, some improved varieties, particularly Bukenge, had significantly lower values (3.10–4.10 g/g). At 100 °C, the pattern reversed (p < 0.001): Mundola (7.50 g/g) and Mulezi (7.13 g/g) had the highest values, whereas Ngorodima decreased to 4.10 g/g. Swelling power depends on water-binding capacity and starch-granule stability during heating (Akonor et al., 2023; Nilusha et al., 2021). The delayed swelling of Mundola and Mulezi at 100 °C may indicate greater initial granule resistance followed by more extensive disruption at high temperature; such responses are influenced by starch organisation and amylose content (Widowati et al., 2025).

Solubility index showed pronounced differences among varieties and temperatures (p < 0.001 at all temperature levels, Figure 2). At 60 °C, Ngorodima was highly soluble (100%), whereas Mulezi and Mundola showed no measurable solubility. At 80–90 °C, several improved varieties became highly soluble, including Mundola (80%), Bukenge (70%), and Mulezi (80% at 90 °C), while Ngorodima remained insoluble through 90 °C. At 100 °C, the values converged partially, although Sikirou, Mundola, and Bukenge reached the highest level (70%). Solubility, therefore, depended strongly on genotype and temperature. The solubility index reflects the release of soluble molecules, particularly amylose, from starch granules during heating (Nilusha et al., 2021). High solubility indicates substantial granule disruption and release of soluble fractions. The elevated values of several improved varieties at 80–90 °C suggest good cooking performance and an ability to form homogeneous pastes, whereas the temporary insolubility of Ngorodima at these temperatures may reflect a more heat-resistant granular structure.

Viscosity differed significantly only at 50 °C (p = 0.0059). Sikirou (10.83 Pa·s), and Bukenge (10.60 Pa·s) had high values comparable with Ngorodima (10.40 Pa·s). At 65–95 °C, varietal differences were no longer significant and viscosity values remained low and relatively similar (approximately 1.77–3.93 Pa·s). Pasting and viscosity depend on starch structure, amylose content, and the capacity of granules to swell and disintegrate during heating (Akonor et al., 2023; Widowati et al., 2025). The high initial viscosity of Sikirou, Bukenge, and Ngorodima may indicate rapid hydration or early release of viscosity-enhancing components, whereas the lack of differences at higher temperatures suggests convergence toward similar rheological behaviour after prolonged heating.

Particle-size distribution differed significantly (p < 0.001) among varieties in every size class, reflecting differences in root texture, cell-wall composition, and milling behaviour. Ngorodima had the finest flour, with high proportions of particles measuring 45–150 μm (56.20%) and <45 μm (27.15%), consistent with its high crude fiber content (2.54%; Table 3). The abundance of fine particles in Ngorodima directly explains its rapid swelling power at intermediate temperatures (60–80 °C; 9.90–10.30 g/g) observed in Section 3.3, as smaller particles have a higher surface-area-to-volume ratio, facilitating faster water penetration and starch granule hydration. Bukenge, by contrast, contained larger coarse fractions (>600 μm, 15.76%; 250–600 μm, 26.10%), indicating a more fibrous or woody cellular structure that resists fine milling. This coarser particle distribution mechanistically explains Bukenge’s lower swelling power at intermediate temperatures (3.10–4.10 g/g), as larger particles present a smaller surface area for water interaction and require longer heating times to achieve full hydration. Sikirou and Mundola were dominated by the intermediate 150–250 μm fraction (38.60 and 35.20%, respectively), representing a balanced particle-size distribution that may confer advantages for both rapid hydration and structural integrity during cooking.

Particle size affects hydration kinetics, dispersibility, cooking rate, texture, and the overall quality of processed products through different mechanisms. Fine particles (Ngorodima) promote rapid water absorption and uniform gelatinization, making them ideal for applications requiring quick cooking and smooth texture (e.g., instant porridges, baby foods) though actual product formulation and safety/acceptability testing are necessary to support such technological hypothesis. Intermediate particles (Sikirou, Mundola) provide a balance between cooking speed and texture retention, making them suitable for traditional preparations (e.g., fufu, high quality cassava flour (HQCF), gari) where both rapid cooking and firm texture are valued. However, that suitability potential of Sikirou and Mundola should be confirmed experimentally in those traditional products. Coarse particles (Bukenge) slow hydration and may cause uneven gelatinization, resulting in a heterogeneous texture with both soft and firm regions—a potentially undesirable characteristic for most cassava-based foods. Akonor et al. (2023) noted that particle-size distribution is a useful indicator of flour suitability for applications requiring uniform dough and fine flour. The fine particles in Ngorodima may promote rapid hydration and water dispersion, whereas the coarser fractions in Bukenge may slow hydration and alter final product texture. Particle-size distribution should, therefore, be considered alongside composition and other functional traits when characterizing improved and local cassava varieties, and milling protocols should be optimized to achieve particle-size distributions appropriate for the intended end-use (Figure 2).

3.4 Multivariate analysis and correlations among cassava root and flour quality traits

Figure 3 illustrates the relationships between cassava varieties and physicochemical, biochemical, and functional characteristics of their fresh roots. Principal component analysis (PCA) showed a clear separation of varieties along the first two components, which together explained 61.0% of the total variance (Dim1, 33.9%; Dim2, 27.1%; Table 5), with PC1 mainly contributed by reducing sugars (9.58%), total sugars (9.08%), and bulk density (8.62%), while PC2 was primarily associated with protein (14.10%), pH (11.91%), and carbohydrate content (11.59%; Table 6). The first component contrasted varieties associated with high carbohydrate and starch contents, together with some functional traits such as swelling power, solubility index, dispersibility, and apparent viscosity, to those characterised by higher protein, titratable acidity (TA), HCN, amylose, bulk density, and water absorption capacity (WAC). Mulezi and Sikirou were positioned on the negative side of Dim1 and were strongly associated with carbohydrates and starch, indicating an energy-dense profile and favourable starch-related technological properties. In contrast, Mundola and Nambiyombiyo were positioned on the positive side of Dim1 and were associated with protein, HCN, titratable acidity, amylose, and density-related attributes, suggesting a different compositional and functional behaviour.

Table 5. Eigenvalues, percentage of explained variance, and cumulative variance of the principal component analysis.

Table 6. Contribution of physicochemical, biochemical, and flour-functional variables to the principal components (%).

TA, titratable acidity; CF, crude fibre; RS, reducing sugars; TS, total sugars; HCN, hydrogen cyanide; WAC, water absorption capacity; LGC, least gelation concentration; BD, bulk density; SP100, swelling power at 100 °C; SI100, solubility index at 100 °C; Visc50, apparent viscosity at 50 °C; Dim., dimension.

The second component primarily discriminated varieties according to the relative contribution of functional and quality-related properties. Bukenge, located in the upper-left quadrant, was more strongly associated with dispersibility, solubility index at 100 °C (SI100), apparent viscosity at 50 °C (Visc50) and, to a lesser extent, lipids, indicating better reconstitution and dispersion behaviour. Mundola, positioned in the upper central-right area, was associated with protein, WAC, bulk density, and amylose, reflecting a profile favourable for water uptake and structural stability in processed products. By contrast, the local variety Nambiyombiyo, located on the positive side of Dim1 but negative on Dim2, was more closely related to pH, crude fibre, moisture, ash, and sugar-related traits, especially total sugars (TS) and reducing sugars (RS). Ngorodima, positioned in the lower central part of the plot, showed a weaker but still noticeable association with pH and moisture-related properties, while remaining distant from the starch-rich group represented by Mulezi and Sikirou (Figure 3).

The orientation of the vectors also provides useful information on interrelationships among traits (Figure 3). The close alignment of carbohydrates and starch indicates positive covariation between these two variables, while the grouping of TS, RS, ash, and moisture vectors had similar orientations, suggesting that these traits tended to vary together among test varieties. Similarly, the proximity of protein, HCN, TA, amylose, bulk density, and WAC indicates that these variables jointly contributed to the separation of Mundola and Nambiyombiyo from the starch-rich varieties. Among the functional traits, dispersibility, solubility index, and apparent viscosity showed similar orientations, suggesting a common contribution to the behaviour of the flours during reconstitution and heating. Overall, the PCA distinguished Mulezi and Sikirou by their high starch and carbohydrate contents, Bukenge by its favourable dispersion and solubility characteristics, Mundola by its association with protein and water absorption, and Nambiyombiyo by its stronger association with sugars, fibre, acidity-related, and safety-related traits, whereas Ngorodima occupied an intermediate position with a tendency toward moisture- and pH-related characteristics.

Pearson correlation analysis revealed significant associations among the physicochemical and biochemical traits of cassava roots (Figure 4). Carbohydrates were strongly and positively correlated with starch (r = 0.857), while starch was strongly correlated with total sugars (r = 0.857*) and reducing sugars with total sugars (r = 0.874), reflecting the predominantly starchy nature of cassava roots (Nilusha et al., 2021; Borku et al., 2025). This pattern was consistent with the PCA, which associated Mulezi and Sikirou with high carbohydrate and starch contents. Ash was positively correlated with reducing sugars (r = 0.632*) and total sugars (r = 0.510), whereas its association with amylose was weak (r = 0.441). Crude fibre was moderately and negatively correlated with amylose (r = −0.483).

Protein was strongly and negatively correlated with carbohydrates (r = −0.908) and starch (r = −0.731), whereas HCN was positively correlated with protein (r = 0.820) and negatively with carbohydrates (r = −0.812) and starch (r = −0.674). HCN was also strongly correlated with titratable acidity (r = 0.905*), which was negatively associated with carbohydrates (r = −0.699) and starch (r = −0.633*), while its association with amylose was weak (r = 0.453). This contrasting profile suggests that Mulezi and Sikirou, characterized by higher starch and lower HCN, may be more suitable for starch-based food applications, whereas Mundola and Bukenge showed relatively stronger associations with protein, acidity, and HCN. Given the food-safety implications of cyanogenic compounds, HCN should be considered alongside nutritional and technological traits in varietal selection (Quinn et al., 2022; Bamidele, 2025). However, these correlations should not be interpreted as causal relationships, and further breeding studies should assess the HCN–starch–protein relationships to ensure that improvements in disease resistance are not accompanied by undesirable changes in food safety or processing quality. Finally, pH was strongly correlated with moisture (r = 0.826), reducing sugars (r = 0.728*), and total sugars (r = 0.602), but showed weak and non-significant associations with amylose (r = 0.059) and HCN (r = −0.075). The close relationships among pH, moisture, and sugars suggest that these traits may jointly influence the physicochemical and processing characteristics of cassava roots. pH is an important indicator of root quality and processing suitability (Babatuyi et al., 2024), while moisture and dry matter influence flour yield, stability, and technological performance (Akonor et al., 2023; Nilusha et al., 2021). Overall, these findings highlight the importance of considering compositional, safety, and functional traits jointly when selecting cassava varieties.

The inclusion of functional properties revealed additional technologically relevant associations. Bulk density was very strongly and negatively correlated with carbohydrates (r = −0.831) and strongly negatively correlated with starch (r = −0.713). Conversely, bulk density was strongly positively associated with HCN (r = 0.741*), protein (r = 0.662), and titratable acidity (r = 0.646). These results indicate that flours derived from carbohydrate- and starch-rich varieties tended to have lower bulk density, whereas varieties associated with protein, acidity, and HCN tended to produce denser flours. Since bulk density influences packaging volume, flour handling, ingredient dosing, and formulation behaviour, these compositional–functional relationships are important for selecting cassava varieties for different processing applications.

Amylose was particularly important for gelation behaviour. A very strong negative correlation was observed between amylose and least gelation concentration (r = −0.809). This means that varieties with higher amylose concentrations tended to require a lower flour concentration to form a gel. From a technological perspective, this relationship is favourable because a lower LGC indicates stronger gel-forming ability at relatively low solids concentration. Amylose was also positively associated with bulk density (r = 0.646) and water absorption capacity (r = 0.554), suggesting that variation in starch molecular composition contributes not only to gel formation but also to flour hydration and packing behaviour (Figure 4). This relationship is consistent with the recognized role of amylose in gel firmness, starch reassociation, and texture development in cassava-based foods.

The swelling behaviour of the flour was also related to the initial physicochemical condition of the roots. Moisture content was strongly and negatively correlated with swelling power at 100 °C (r = −0.796*). Thus, varieties with higher fresh-root moisture tended to produce flours with lower swelling capacity at high temperatures. This observation suggests that the dry-matter status of cassava roots may be related to the ability of starch granules to absorb water and expand during heating. Because swelling power is an important indicator of starch hydration, cooking behaviour, and the consistency of cassava-based pastes, the relationship further demonstrates the importance of considering fresh-root dry matter together with flour functionality when selecting varieties for processing.

The solubility-related traits also showed notable relationships with root biochemical composition. The solubility index at 100 °C was very strongly and negatively correlated with reducing sugars (r = −0.881) and strongly negatively correlated with total sugars (r = −0.765). It was also strongly negatively associated with pH (r = −0.787*). In contrast, SI100 was positively associated with dispersibility (r = 0.627). This latter relationship suggests that flours releasing a greater soluble fraction during high-temperature treatment also tended to disperse more readily in water. The negative associations with soluble sugars indicate that varieties richer in reducing and total sugars did not necessarily exhibit greater starch-related solubility at 100 °C, highlighting the distinction between pre-existing soluble carbohydrates and starch components released during heating. Dispersibility itself was strongly and negatively correlated with reducing sugars (r = −0.669), while its association with SI100 was positive (r = 0.627). This implies that the ability of cassava flour particles to disperse in water depends on the structural characteristics of the flour rather than simply on the concentration of soluble sugars present in the roots. Such behaviour may affect flour reconstitution, lump formation, hydration uniformity and the preparation of porridges, pastes and other cassava-based foods.

Several relationships involving pH and moisture further linked fresh-root properties to subsequent processing behaviour. pH was very strongly positively associated with ash (r = 0.826), and strongly associated with reducing sugars (r = 0.728) and total sugars (r = 0.602). However, pH was strongly negatively related to SI100 (r = −0.787). Moisture was strongly positively associated with total sugars (r = 0.711) and reducing sugars (r = 0.655), while its strong negative association with SP100 indicates that more succulent roots were generally linked to reduced flour swelling at high temperature. These relationships suggest that fresh-root biochemical status may influence subsequent flour functionality and should, therefore, not be considered independently from processing performance.

Collectively, the integrated correlation analysis demonstrates that composition, food safety, and functional performance are closely interrelated in the cassava varieties evaluated. Mulezi and Sikirou, characterised by comparatively high carbohydrate and starch contents, were associated with a biochemical profile favourable for starch-based processing and relatively lower HCN concentrations. In contrast, Mundola and Bukenge were more closely linked with protein, acidity, HCN, and density-related characteristics, while the local varieties Nambiyombiyo and Ngorodima showed stronger associations with moisture, sugars, and other compositional attributes. The results, therefore, show that varietal evaluation should not be based solely on agronomic yield or individual nutritional components. Rather, cassava breeding and selection programmes should simultaneously consider dry matter, starch, cyanogenic potential, amylose composition, hydration, gelation, swelling, solubility, dispersibility, and bulk-density characteristics to identify genotypes that combine food safety with desirable processing quality.

Because the correlations describe statistical co-variation rather than causal mechanisms, the observed relationships should be interpreted as associations among traits within the test varieties. Nevertheless, their consistency with the integrated PCA provides complementary evidence that physicochemical, biochemical composition, and flour functionality jointly contribute to the technological differentiation of cassava genotypes. Collectively, the correlations confirm that composition, safety, and functionality should be evaluated jointly when selecting cassava varieties, and that scientific understanding of the biochemical trade-offs underlying varietal differences is essential for rational breeding and processing strategy development.

4 Conclusion

This study compared the physicochemical, functional, and technological performance of CBSD- resistant cassava varieties with that of locally preferred varieties. Findings showed that CBSD-resistant cassava varieties exhibit distinct combinations of food-safety, starch-composition and functional traits, some of which compare favourably with locally preferred varieties while others reveal clear breeding trade-offs. Among the improved varieties, Mulezi and Sikirou showed the most favourable overall performance across physicochemical, functional, and processing-related traits. Their starch content and processing behaviour were comparable with or superior to those of the best local cassava varieties. Bukenge, however, combined relatively high HCN with weaker technological performance, which may constrain its use in traditional processing unless effective detoxification and appropriate processing protocols are applied. Agronomic performance alone is, therefore, insufficient as a basis for recommending improved cassava varieties. Processing quality, food safety, functional performance, and consumer acceptance are equally important determinants of successful integration into local food systems.

Mulezi and Sikirou are ready for wider dissemination to a recommendation for further multi-location, processing, and consumer validation. Further studies should confirm these findings across locations, seasons, and processing methods, and evaluate sensory acceptance using larger and more diverse consumer panels.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Wasiu Awoyale, Kwara State University, Nigeria

Reviewed by: Michael Bayode, Federal University of Technology, Nigeria

Saurabh Dilip Bhandare, Laboratorium Scientiae et Studiorum Investigativorum, India