Exploring the nutritional composition, physicochemical properties, and biological characteristics of mature banana varieties (Musaceae)

Abstract

Despite the global significance of bananas, detailed characterization of their nutritional, physicochemical, and bioactive properties remains limited, especially for rare or regionally cultivated varieties. This study profiles 13-banana cultivars from Vietnam to identify varietal differences and assess their potential in food applications, using a multi-dimensional framework including morphology, surface color, and texture. Physicochemical characteristics and proximate nutritional components were assessed using validated methodologies. Moreover, key bioactive compounds and antioxidant activities were quantified via spectrophotometric assays. Results revealed significant inter-cultivar variation across all analytical domains. Ta Qua exhibited the largest fruit morphology and the highest carbohydrate content (>40.29 g/100g), while Cha Bot demonstrated elevated concentrations of phenolic and flavonoid compounds, correlating with robust antioxidant activity (1.17–1.67 mgAA/gDW). Conversely, Sap Trang presented limited antioxidant capacity but was comparatively rich in protein (1.89 g/100g). The study highlights the importance of cultivar-specific profiling to support evidence-based selection for nutrition, breeding, and food development.

Highlights

• •The physicochemical, nutritional, antioxidant properties of 13 banana varieties were explored.
• •Cha Bot banana was the most promising with high polyphenols, flavonoids, and antioxidants.
• •Correlation, PCA, and dendrogram were used to analyze the characteristics of the banana varieties.
• •Seven texture indices for each banana variety were comprehensively explored.

1. Introduction

Bananas (Musa spp.), among the most consumed fruits globally, including numerous cultivated varieties and hundreds of wild species (Anyasi et al., 2015; Heslop-Harrison & Schwarzacher, 2007). However, all of them are derived from two wild species, Musa acuminata and Musa balbisiana, with genomes ranging from diploid to tetraploid (Emaga et al., 2007). They serve as staple crops in tropical and subtropical regions, offering essential nutrients such as potassium, dietary fiber, and bioactive compounds (Anyasi et al., 2013; Awedem et al., 2015). Currently, bananas rank as the fourth most important food crop worldwide, with production increasing from 69 million tons (2000–2002) to over 100 million tons (2017–2019), valued at approximately USD 31 billion (Hang et al., 2018; Lamessa, 2021).

Banana cultivars exhibit considerable variation in nutritional and physicochemical properties. In South India, protein content among eight varieties (Palayankodan, Rasakadali, Robusta, Poovan, Nendran, Kadali, Red banana, and Padatti) ranged from 0.90 to 1.37 g/100g, while total soluble solids (TSS) varied between 17.83 and 23.90 °Bx (Siji & Nandini, 2017). Similar diversity in carbohydrate (13.46 – 26.46 g/100g), protein (up to 1.37 g/100g), and acidity content (up to 1.28 g/100g) has been reported in Sri Lanka (Nadeeshani et al., 2021). Moreover, environmental factors – such as soil composition, temperature, and cultivation methods – have been shown to significantly influence fruit composition (Thinh et al., 2021).

Bananas are rich in antioxidants and other vital compounds, including essential minerals, carbohydrates (Khawas et al., 2014), vitamin E and C, phenolics, flavonoids, tannins, and carotenoids (Arora et al., 2008), which play roles in oxidative stress reduction (Sad et al., 2017), gut health, and disease prevention (cancer and cardiovascular) (Singh et al., 2016). The reported variation in β-carotene and α-carotene across Musa genotypes further emphasizes the need for cultivar-specific profiling (Arora et al., 2008; Sad et al., 2017). Yet, many published studies tend to focus on isolated components or commercial cultivars, often evaluating them under differing environmental conditions. Without conducting multi-varietal analysis within a single locality, it becomes difficult to disentangle genetic effects from environmental influences – thereby limiting a comprehensive understanding of the true nutritional and functional potential of banana germplasm.

In Vietnam, over 30 banana cultivars have been found, including both introduced and locally bred varieties (Cruz-Rosero et al., 2017). However, most research has focused on a limited number of commercial types, such as Xiem, Dole, and Cau. Dinh & Doan, 2024 reported significant variation in folic acid content and peel brightness across five Vietnamese varieties (Xiem, Cau, Gia Huong, Dole, and Com). Besides, the Tieu variety in Northern Vietnam showed higher protein and lipid content (2.53 and 5.00 g/100g, respectively) compared to its Taiwanese counterpart (0.36 and 5.73 g/100g), highlighting the role of local agro-ecological factors (Thinh et al., 2021). Despite these insights, comprehensive comparative profiling-encompassing nutritional, physicochemical, and bioactive characteristics-remains lacking for lesser-known cultivars such as Ta Qua, Laba, Cha Bot, Sap Vang, Hot, Ngu, Mat, Lun, Tieu, Sap Trang or Gia Huong. Even for previously reported varieties, a complete and systematic evaluation is still unavailable.

Therefore, this study aims to comprehensively evaluate 13 banana cultivars cultivated in Vietnam, focusing on five nutritional components, four bioactive properties, and seven physicochemical parameters. Advanced statistical tools – including hierarchical clustering, principal component analysis (PCA), and correlation analysis – will be used to identify compositional patterns and classify cultivars based on their similarities. The hypothesis underpinning this study is that genetic diversity and localized cultivation conditions significantly contribute to the observed variations between banana cultivars grown in Vietnam and those cultivated elsewhere. The resulting findings are anticipated to provide a scientific foundation for informed applications in food systems, public health, and agricultural development – emphasizing the unique strengths of each cultivar rather than advocating for a single superior type.

2. Material and methods

2.1. Plant material

The banana varieties Xiem, Gia Huong, Hot, Mat, Tieu, and Cau were harvested in Dong Nai (11°03’16”N, 107°08’49”E) 90 days after flowering. Sap Vang and Sap Trang bananas were harvested 120 and 150 days after flowering, respectively, in Ben Tre (10°14’54”N, 106°22’34”E). Ta Qua bananas were harvested in Tra Vinh (9°48’50”N, 106°17’56”E) 180 days after flowering. Laba and Tieu bananas were harvested in Lam Dong (11°41’04”N, 108°08’28”E) 120 days after flowering. Lun bananas were harvested in Binh Dinh (14°11’43”N, 109°52’48”E) 150 days after flowering, and Ngu bananas were harvested in Dong Thap (10°34’32”N, 105°40’59”E) 90 days after flowering. All 13 banana varieties were selected at the appropriate harvest maturity for fresh consumption, in accordance with TCVN 1872:2019 on fresh bananas, and were transferred to the laboratory of the Institute of Applied Technology and Sustainable Development. Samples were analyzed immediately after collection or stored at a temperature of 20 ± 2°C for a maximum of one day.

2.2. Chemicals and reagents

Sodium carbonate (99.8%), Sodium hydroxide (> 96%), Aluminum chloride (97%), Potassium sodium tartrate tetrahydrate (99%), Potassium acetate (98.14%), Ascorbic acid (99.7%), 2,2’ – Azino – Bis – 3 –Ethylbenzothiazoline – 6 – Sulfonic Acid (ABTS) (> 98%, Merck), 2,2 – Diphenyl – 1–1 picrylhydrazyl (DPPH) (95%), Folin – Ciocalteu (FCR), Ethanol (90%), Potassium iode (≥ 99.5%), Potassium iodate (≥ 99.7%), Sulfuric acid (≥ 98%), Acetate acid (37%), Aluminum chloride (97%), Fomaldehyde (99.8%), Ascorbic acid (99.7%), Sodium phosphate monobasic dihydrate (99%), Lead (II) acetate basic (99.5%), Copper (II) sulfate pentahydrate (99%), 3.5-Dinitro salicylic acid (DNS), Diethyl ether (98%), Phenolphthalein were purchased from Merck (Germany), Starch (99%), and Acetone (99.5%) were purchased from Xilong (China).

2.3. Physical characteristics

2.3.1. Moisture content

Moisture content (g/100gFW) was determined based on the weight difference before and after heating at 105°C until a constant weight was achieved. A moisture analyzer (MB90, Ohaus, USA) was used to assist in determining the moisture content of each 1 g sample with 5 repetitions (AOAC 925.10, 1990; Vu et al., 2023).

2.3.2. Color

The color measurement method used the CIE Lab color space, with three values L*, a*, and b* serving as the reference for color. The CR400 device (Minolta, Japan) was used to collect the values L* (lightness 0-100), a* (green to red), and b* (blue to yellow). The device was calibrated using a standard plate to determine the color values and was placed perpendicular to the samples of the flesh and peel of each banana, which were arranged on a table. The results were recorded on the device’s display for each button press, capturing readings from 5 different positions (Vu et al., 2025).

2.3.3. Texture

The texture measurement method for the banana samples was carried out using a Texture Profile Analysis (TPA) with the Texture Analyzer (TA.XTplusC.0001, Stable Micro System Ltd., UK), based on the parameters extracted from the force-time curve obtained. Bananas with similar ripeness levels were selected, and the peels were removed. The flesh was then cut into equal-sized shapes (with a length and radius of 2 cm). A 75 mm flat cylindrical probe (P/75) was used. The setup parameters included pre-test speed (1 mm/s), test speed (5 mm/s), post-test speed (5 mm/s), deformation (20%), test duration (2.5 s), and applied force (5 g) (Bourne, 1990).

2.3.4. Morphological characteristics

The length and weight of the fruit were determined using a measuring tape and a specialized weighing device. The measurements were repeated 20 times (n = 20).

2.4. Chemical properties

2.4.1. Ash

Ash content (g/100gFW) was determined based on the weight difference of the sample before and after ignition at high temperature. Approximately 25 ± 5 g of the sample was ignited at a temperature of 550 – 600 °C for 6 h (HTC08/14, Narberthem, Germany) (AOAC 940.26, 1990; Vu et al., 2024).

2.4.2. Protein

The protein content of the banana varieties (g/100gFW) was determined using the Lowry method. Reagent A was prepared by sequentially adding 2 g of sodium carbonate and 100 mL of sodium hydroxide 0.1 N. Reagent B was prepared by sequentially adding 0.5 g of copper (II) sulfate pentahydrate to 100 mL of potassium sodium tartrate tetrahydrate (10 g/L). Reagent C was a combination of reagents A and B in a 5:1 (v/v) ratio. A total of 5 g of the sample was hydrolyzed with 100 mL of 2 N sodium hydroxide to obtain the sample solution. For 0.1 mL of the sample solution, 0.4 mL of reagent C was added and allowed to stabilize for 10 min. Then, an additional 0.5 mL of Folin-Ciocalteu reagent was added, mixed using a Vortex device, and incubated in the dark for about 30 min. After incubation, the absorbance was determined at a wavelength of 660 nm with the aid of a UV-VIS spectrometer (UV-11, Spectro, USA). Protein content was calculated based on a standard curve of bovine serum albumin: y = 0.1220x – 0.0003 (R² = 0.9992), using concentrations ranging from 0 to 1 g/L (Lowry et al., 1951; Majaliwa et al., 2021).

2.4.3. Lipid

The lipid content (g/100gFW) was determined based on the principle of weight difference after extraction (Soxhlet method). A dried sample quantity of 5 g was used for analysis. The extraction was carried out at 80 ± 2°C for 8 h (AOAC 954.02, 1990; Santos et al., 2021).

2.4.4. Crude fiber

The crude fiber content (g/100gFW) was determined based on the principle of weight difference after combustion of the sample that was hydrolyzed with acid and base. An amount of 10 g of the sample was sequentially hydrolyzed with 50 mL of 12.5 g/L sulfuric acid, followed by 50 mL of 12.5 g/L sodium hydroxide. The solid residue obtained after hydrolysis was incinerated at 550 – 600 °C for 7 h (AOAC 962.09, 1990; Vu et al., 2024).

2.4.5. Carbohydrate

The carbohydrate content (g/100gFW) was determined based on the components of protein, ash, fat, and moisture content of the sample (U.S. Food and Drug Administration, 2013).

$$\text{Carbohydrate}=100–\left(\text{Protein}+\mathrm{Ash}+\text{Lipid}+\text{Moisture}\right)$$

2.4.6. Titrated acidity

The titratable acidity (g/100gFW) was determined based on the principle of acid-base titration. A mixture of 10 g of the sample and 150 mL of distilled water was prepared for extraction. Approximately 25 mL of the extract was added with 3 drops of (1 g/L) phenolphthalein indicator and titrated with 0.1 N sodium hydroxide solution until a stable pink color appeared for 30 s (AOAC 942.15, 1990; Thuy et al., 2024).

2.4.7. pH

The pH value was determined based on the activity of H₃O⁺ ions in the solution. The pH value of each 20 g sample was determined using a pH meter (HI2211, Hanna, China) at 22 ± 1 °C (AOAC 973.41, 1990; Thuy et al., 2024).

2.4.8. Reducing sugar

The reducing sugar content (g/100gFW) was determined based on the reaction of reducing sugars with 3.5-Dinitrosalicylic acid (DNS) reagent in an alkaline environment to form an orange-red complex. Total 2 mL of the extract solution was combined with 1 mL of DNS reagent and incubated in a water bath at 95°C for 5 min. The absorbance of the samples was determined at 534 nm using a UV-VIS spectrophotometer (UV-11, Spectro, USA), and the reducing sugar content was calculated based on a glucose standard curve: y = 0.2441x + 2.9172 (R² = 0.9992), using concentrations ranging from 0 to 1 g/L (AOAC 925.36, 1990; Pham et al., 2024a).

2.4.9. Total soluble solids

The total soluble solid (°Bx) was determined based on the percentage concentration of soluble solids in the solution. Approximately 20 ± 1 g of the sample was pureed and placed on the prism of an electronic refractometer (WM-7, Atago, Japan) (AOAC 932.12, 1990; Sinanoglou et al., 2023).

2.5. Bioactive components

2.5.1. Total phenolic content

The total phenolic content (mgGAE/gDW) was determined based on the phosphor-tungsten and phosphor-molybdate complexes in the Folin-Ciocalteu reagent. About 5 g sample was extracted with 50 mL of 96% ethanol. Approximately 0.1 mL of the extract, 0.5 mL of (100 mL/L) Folin-Ciocalteu reagent and 0.4 mL of (75 g/L) sodium carbonate solution were added and homogenized using a Vortex device. The mixture was then incubated in the dark for 60 min. The absorbance of the samples was determined at a wavelength of 765 nm using a UV-VIS spectrophotometer (UV-11, Spectro, USA), based on a gallic acid standard curve: y = 127.2584$x$ – 17.3250 (R² = 0.9994), using concentrations ranging from 0 to 40 mg/L (Pham et al., 2024b).

2.5.2. Total flavonoid content

The total flavonoid content (mgQE/gDW) was determined based on the principle of color formation with AlCl₃ in an alkaline environment. Extraction was carried out by mixing 5 g of the sample with 50 mL of 96% ethanol. Following this, 0.5 mL of the extract, 4.3 mL of 96% ethanol, 0.1 mL of (100 g/L) aluminum chloride, and 0.1 mL of 1M sodium carbonate were added and homogenized using a Vortex device. The mixture was then incubated in the dark for 30 min. The absorbance of the solution was determined at a wavelength of 415 nm using a UV-VIS spectrophotometer (UV-11, Spectro, USA), based on a Quercitrin standard curve: y = 178.8668x + 1.7725 (R² = 0.9990), using concentrations ranging from 0 to 40 mg/L (Pham et al., 2024b).

2.5.3. Total ascorbic acid

The ascorbic acid content (mg/100gDW) was determined based on the principle of oxidation reduction for samples containing ascorbic acid. A drop of excess iodine solution, when meeting starch, forms a blue-black complex. The iodine solution is a mixture of 5 g potassium iodide, 0.268 g potassium iodate, 30 mL of 3 M sulfuric acid, and was made up to 500 mL with distilled water. Approximately 30 g of the sample was extracted with 50 mL of distilled water. Titration was performed on 20 mL of the extract by adding 10 drops of (10 g/L) starch indicator, followed by iodine solution until a stable blue endpoint lasting 30 s was achieved. The titration of the standard ascorbic acid (0.1 g/mL) was performed similarly, but the sample was replaced with the standard ascorbic acid solution (Mohammed & Yousef, 2024).

2.5.4. Total carotenoid content

The total carotenoid content (mg/100gDW) was determined based on the absorbance of chlorophyll a, chlorophyll b, and total carotenoids. One gram of sample was homogenized with 10 mL of an acetone – water mixture (4:1, v/v) for 2 min. Ultrasound was applied for 3 min in an ice bath to prevent overheating. Samples were then centrifuged at 5,000 rpm for 10 min at 20 °C. Absorbance was measured at 663.6 nm (chlorophyll a), 646.6 nm (chlorophyll b), and 470.0 nm (total carotenoids) (Andre., 1976).

2.6. Antioxidant activity

2.6.1. ABTS antioxidant activity

The ABTS antioxidant activity (mgAA/gDW) was determined based on the reaction of ABTS with oxidizing compounds. ABTS was reduced from a dark green color to a colorless form when an antioxidant compound was added. Total 5 g of the sample was ground with 50 mL of 96% ethanol and filtered through filter paper to obtain the extract. Subsequently, 0.5 mL of the extract, 1.5 mL of ABTS solution was added, homogenized using a Vortex device, and incubated in the dark for 30 min. The sample was measured for maximum absorbance at a wavelength of 734 nm with the aid of a UV-VIS spectrophotometer (UV-11, Spectro, USA), based on a standard curve equivalent to ascorbic acid: y = -22.5216x + 20.1603 (R² = 0.9990), using concentrations ranging from 0 to 40 mg/L (Do et al., 2024).

2.6.2. DPPH antioxidant activity

The DPPH antioxidant activity (mgAA/gDW) was determined based on the ability of antioxidants to absorb hydrogen molecules. Total 5g of the sample was ground with 50 mL of 96% ethanol and filtered through filter paper to obtain the extract. The extract (0.5 mL) was reacted with 1.5 mL of DPPH solution to initiate the assay. The mixture was homogenized using a Vortex device and incubated in the dark for 30 min. The sample was measured for maximum absorbance at a wavelength of 517 nm with the aid of a UV-VIS spectrophotometer (UV-11, Spectro, USA), based on a standard curve equivalent to ascorbic acid: y = -36.7509x + 28.3107 (R² = 0.9995), using concentrations ranging from 0 to 40 mg/L (Pham et al., 2024b).

2.7. Statistical and data analysis

The results were presented as the mean ± standard deviation (with at least three repetitions). JMP Pro 18 software (SAS Institute, Cary, North Carolina, USA) was used for statistical analysis (95% confidence level). One-way ANOVA was used to compare the means of different banana cultivars. Tukey HSD was used to test for significant differences.

3. Result and discussion

3.1. Morphological characteristics

Morphological characteristics such as fruit length and weight were used as criteria to describe and distinguish different fruit types and varieties. In the same fruit type, morphological differences are mainly due to the variation between varieties (Ayala-Silva et al., 2009). The morphological evaluation results of 13 banana varieties harvested in Vietnam showed significant differences in the fruit length and weight values (Table 1).

Table 1.

Length and weight of 13 mature banana varieties

Varieties	Common names	Genomic	Reference	Abbre.	Length (cm)	Weight (g)
Xiem	Saba, Kluai Nam Wa, Pisang kepok, Pisang abu	ABB	(Reginio et al., 2020)	XI	8.68 ± 0.25f	43.92 ± 0.40g
Gia Huong	Cavendish	AAA	(Tran et al., 2016)	GH	13.35 ± 0.49d	98.31 ± 0.35d
Cha Bot	Pitogo, Kluai Nam Khom, Pisang Kastroli	ABB	(Valmayor et al., 2000)	CB	10.26 ± 0.35e	60.92 ± 0.52f
Sap Vang	Sap nghe banana	-	-	SV	7.75 ± 0.07gh	36.26 ± 0.32j
Hot	Pisang Batu, Butuan	BB	(Assani et al., 2002)	HO	10.77 ± 0.04e	40.58 ± 0.83h
Laba	Pisang Berangan,	AAA	(Ting et al., 2009)	LB	14.25 ± 0.35c	98.58 ± 0.60d
Ta Qua	Pisang Tanduk,	AAB	(Kemal et al., 2018)	TQ	35.83 ± 0.53a	292.11 ± 0.20a
Ngu	Latundan, Pisang Mas, Pisang Susu	AA	(Megia, 2005)	NG	14.20 ± 0.14c	128.64 ± 0.69c
Mat	Cardaba, Pisang Kepok Besar	BBB	(Valmayor et al., 2000)	MA	8.50 ± 0.14fg	39.83 ± 0.81hi
Lun	Dwarf Cavendish	AAA	(Valmayor et al., 2000)	LU	14.54 ± 0.62bc	83.91 ± 0.64e
Tieu	Jari Buaya	AA	(Valmayor et al., 2000)	TI	15.31 ± 0.13b	146.31 ± 0.95b
Cau	Cau man, Pisang Masan, Jarum, Veinte Cohol	AA	(Valmayor et al., 2000)	CA	7.30 ± 0.42h	40.20 ± 0.28hi
Sap Trang	Pisang Kepok Kuning	BBB	(Valmayor et al., 2000)	ST	8.50 ± 0.57fg	39.00 ± 0.10i

Note: Different letter groups in the same indicator and analysis object show statistically significant differences (p < 0.05). Abbre: Abbreviations; (-): Not determined

The ANOVA results revealed that the Cau banana had the shortest length (7.30 cm) among the 13 banana varieties assessed (p < 0.05). In contrast, the length of the Ta Qua banana was exhibited the greatest fruit length at 35.83 cm, making it the largest banana variety observed in this study. These findings are consistent with previous reports on the length of the Embul, Seeni, and Kolikuttu banana varieties harvested in Sri Lanka, with lengths of 10.5, 10.5, and 14.3 cm, respectively (Wasala et al., 2015), and the Tieu xanh banana variety grown at the Dinesh farm (India, 15.42 cm) (Soltani et al., 2011). The differences in fruit size were also reflected in fruit weight, with weights ranging from 36.26 to 292.11 g. Notably, the Ta Qua banana had the highest fruit weight (292.11 g), while the Sap Vang banana recorded the lowest (36.26 g), with both values showing statistically significant differences from the remaining 11 varieties (p < 0.05). These results suggest a positive correlation between fruit length and weight. Furthermore, the length values recorded for Vietnamese banana varieties such as Xiem, Gia Huong, Cha Bot, Hot, LaBa, Ngu, Lun, and Tieu are comparable to international varieties such as Mabonde, Muomva-red, and Luvhele, which typically range from 9 to 16 cm (Anyasi et al., 2015). In addition, the size range of all 13 varieties in this study aligns with that of the Cavendish banana from Malaysia (6–18 cm), with the exception of the Ta Qua variety, which exceeds this range (Alkarkhi et al., 2011).

3.2. Appearance and color of 13 banana varieties

Color is one of the most critical sensory attributes influencing consumers’ initial perception and acceptance of food products. In bananas, the color of both the peel and flesh serves as a visual indicator of ripeness, freshness, and overall quality, directly affecting consumer preference and purchasing behavior. Differences in color can also suggest variations in pigment composition, maturity stage, or cultivar-specific traits, which are relevant for both marketing and processing purposes.

The results for the values of L*, a*, and b* were more clearly demonstrated by the visual images of the tested banana varieties (Supplementary Figure 1). The comparison of the color parameters of the 13 banana varieties also revealed significant statistical differences between them (p < 0.05) (Figure 1). The color analysis results of the banana peel showed that the L* values of the banana varieties ranged from 56.60 to 76.48. This result is similar to the two banana varieties, Agbagba and Obino L'e wai, which had L* values ranging from 60.96 to 84.86 (Singh & Kulkarni, 2006). Among them, the Cau banana had the highest lightness (76.48). However, the Hot banana had a very low L* value (47.52), which could be due to its higher anthocyanin content compared to the other 12 varieties. A previous study reported that anthocyanin content has an inverse correlation with L*, meaning the higher the anthocyanin content, the darker the color, and the lower the L* value (Ove et al., 2019). The varieties Gia Huong, LaBa, and Tieu were observed to have a greenish color with a* and b* values of -1.18/32.16, -2.37/39.51, and -0.72/41.19, respectively. A more distinct green color was found in the Lun banana variety, with a* value of -10.78 and b* value of 31.06. Additionally, the Cavendish banana (Musa cavendishi, subgroup AAA) was found to retain its green color when ripe due to the incomplete breakdown of chlorophyll (Janave & Sharma, 2004). Most banana varieties showed a tendency towards yellow, with a* values ranging from 1.09 to 6.33 and b* values ranging from 16.47 to 43.38. Previous reports also indicated similar colors with a* values ranging from 3.56 to 5.28 and b* values from 37.62 to 60.03 (Al-Dairi & Pathare, 2024).

Figure 1.

L*, a*, b* values of the flesh and peel of different banana varieties. Different letters within the same parameter and sample indicate a statistically significant difference (p < 0.05). L*: lightness (0 = black, 100 = white); a*: +60 (red) to -60 (green); b*: +60 (blue) to -60 (yellow).

Although the peel colors varied, the flesh exhibited low variation, with L* values ranging from 69.91 to 76.48, a* values from -2.28 to 2.30, and b* values from 14.07 to 30.79. The analysis of a* and b* values indicated that the flesh of the banana varieties tended toward yellow. This yellow hue is not merely associated with the ripening stage but is also intrinsic to the fruit (Singh & Kulkarni, 2006). Similar results were found for six banana varieties, Parameters, Aloga, Aloga 2M, Gnivlan, Kpahissi, Orishele, and Pelipita, with a* values from 3.92 to 9.77 and b* values from 15.91 to 25.84 (Honfo et al., 2020). Although these results differ from the current study, they still reflect the characteristic yellow color of the fruit. Overall, the variation in yellow color across these varieties may be related to changes in carotenoid content in the fruit (Bugaud et al., 2007). A previous report indicated that the a* and b* values did not show a consistent trend with each other, and one factor affecting the color analysis of the fruit peel could be the presence of dispersed black spots around the fruit (Bin Ramli et al., 2010).

3.3. Texture profiles of banana varieties

Texture is an important parameter for identifying the physical characteristics of fruit. The values for hardness, adhesiveness, elasticity, cohesiveness, gumminess, chewiness, and resilience were determined to clearly provide the characteristics of each banana variety. The texture profile analysis (TPA) results of banana flesh was presented in Table 2.

Table 2.

Texture profiles of banana varieties

Varieties	Hardness (g)	Adhesiveness (g.s)	Springiness	Cohesiveness	Gumminess	Chewiness	Resilience
XI	753.46 ± 265.53d	-46.15 ± 11.91bc	0.80 ± 0.14a	0.68 ± 0.06abc	504.57 ± 142.64def	392.62 ± 71.66bc	0.69 ± 0.15a
GH	1704.36 ± 320.13abc	-54.45 ± 18.01bc	0.55 ± 0.05a	0.53 ± 0.02ef	898.42 ± 186.00bcd	499.28 ± 139.29b	0.42 ± 0.03bcd
CB	1136.26 ± 124.45bcd	-47.00 ± 6.45bc	0.74 ± 0.04a	0.57 ± 0.02abcde	645.25 ± 50.06bcde	480.95 ± 66.25bc	0.53 ± 0.03abcd
SV	802.44 ± 221.74d	-51.61 ± 12.49bc	0.68 ± 0.08a	0.72 ± 0.05a	570.16 ± 144.85cdef	389.48 ± 111.76bc	0.53 ± 0.01abcd
HO	1667.18 ± 30.11abc	-41.18 ± 14.87abc	0.61 ± 0.05a	0.49 ± 0.02ef	812.36 ± 48.46bcde	495.09 ± 139.29b	0.38 ± 0.03cd
LB	967.43 ± 34.92cd	-51.24 ± 15.38bc	0.62 ± 0.08a	0.48 ± 0.02ef	462.00 ± 28.45ef	285.39 ± 50.03bc	0.32 ± 0.04d
TQ	2488.80 ± 353.89a	-45.25 ± 16.33d	0.68 ± 0.09a	0.68 ± 0.06abc	1688.00 ± 16.33a	1160.81 ± 289.55a	0.65 ± 0.12ab
NG	1201.01 ± 208.74bcd	-55.59 ± 24.23bc	0.72 ± 0.26a	0.70 ± 0.08ab	840.05 ± 189.74bcde	570.79 ± 94.95b	0.63 ± 0.21abc
MA	1788.87 ± 195.79abc	-30.00 ± 0.14ab	0.60 ± 0.05a	0.55 ± 0.03bcdef	971.96 ± 70.47bc	585.98 ± 365.33b	0.50 ± 0.02abcd
LU	453.17 ± 122.12d	-0.34 ± 0.23a	0.54 ± 0.11a	0.51 ± 0.09def	224.82 ± 25.88f	120.56 ± 9.16c	0.41 ± 0.06bcd
TI	1917.62 ± 597.83ab	-71.50 ± 14.80c	0.61 ± 0.03a	0.54 ± 0.08cdef	995.92 ± 166.52b	603.55 ± 93.48b	0.41 ± 0.04bcd
CA	1843.92 ± 434.03ab	-46.66 ± 14.68bc	0.61 ± 0.02a	0.40 ± 0.04f	724.38 ± 125.81bcde	441.34 ± 84.85bc	0.29 ± 0.05d
ST	964.17 ± 298.30cd	-39.76 ± 11.13abc	0.82 ± 0.22a	0.67 ± 0.06abcd	629.84 ± 313.29cdef	529.21 ± 222.47b	0.51 ± 0.09abcd

Note: Different letter groups in the same indicator and analysis object show statistically significant differences (p < 0.05).

The ANOVA results indicated that the hardness of the Ta Qua banana (2488.80 g) and the Lun banana (453.17 g) was the highest and lowest, respectively, compared to the other varieties. A study has shown that the hardness of bananas is influenced by moisture content, and further analysis indicated a significant increase in ROS levels, accompanied by enhanced lipid peroxidation processes, which substantially reduced starch content and disrupted cellular integrity (Wang et al., 2024). This means that when the moisture content in bananas is high, the fruit flesh tends to be softer, and vice versa, which is consistent with the findings of this study. Additionally, the gumminess (1687.99 g) and chewiness (1160.81 g) of this variety were also found to be significantly higher (p < 0.05) compared to the other 12 banana varieties. However, the remaining 12 varieties showed no significant statistical difference (p > 0.05). A previous report indicated that the adhesiveness of bananas depends on the variety (Marta et al., 2019). The formation of high-adhesiveness characteristics may be influenced by the pectin content in bananas (Wang et al., 2018), or by Ethrel, which activates enzymes such as pectinase, cellulase, and hemicellulase that break down cell structures, increasing the softness and adhesiveness of bananas (Bugaud et al., 2013). Specifically, the Mat banana variety (-30.00 g.s) was found to have the highest adhesiveness value, while the Tieu banana had the lowest adhesiveness (-71.50 g.s). A previous report indicated a positive correlation between pectin content and the adhesiveness of fruit (Wang et al., 2018) and also revealed that pectin plays a key role in softening the structure in bananas, leading to increased cohesiveness (Prabha & Bhagyalakshmi, 1998). Moreover, in the elasticity test, the three banana varieties Sap Trang, Xiem, and Cha Bot had the highest results at 0.82, 0.80, and 0.74, respectively, with the lowest being the Hot banana (0.54). Overall, physiological and biochemical changes such as the breakdown of pectin, cellulose, hemicellulose, and other polysaccharides into cells by enzymes, the movement of water from the skin, and the hydrolysis of starch into sugars may lead to structural differences in bananas (Ward et al., 1989).

3.4. Basic components of banana varieties

Analyzing basic indicators like moisture, carbohydrate, fiber, protein, fat, and ash helps assess the processing suitability and nutritional value of banana cultivars. Understanding these traits supports efficient use of each cultivar, reduces postharvest losses, preserves biodiversity, and promotes the development of diverse, value-added banana products.

The differences between the banana varieties showed significant differences in the basic nutritional components, including protein, lipid, moisture, fiber, and carbohydrate content (p < 0.05) (Table 3). The moisture content results showed that the banana varieties had moisture levels ranging from 53.25 to 72.39 g/100g. A study by Nadeeshani et al. (2021) reported that five banana varieties (Seeni Vaazhai, Puli Vasszha, Kolikuttu, Seevvaathai, and Poo Vaazha) had moisture contents ranging from 67.24 to 73.14 g/100g. This result is similar to three banana varieties in this study – Ngu (69.74 g/100g), Mat (68.32 g/100g), and Lun (72.39 g/100g). The ash content of the 13 Vietnamese banana varieties ranged from 0.77 to 1.20 g/100g, which is relatively lower than that of four banana varieties (Silk, Mysore, Gros Michel, and Pisang Awak) (Hettiaratchi et al., 2011).

Table 3.

The basic components of banana varieties were determined and expressed on a fresh weight basis

Banana varieties	Moisture (g/100g)	Ash (g/100g)	Protein (g/100g)	Fat (g/100g)	Fiber (g/100g)	Carbohydrate (g/100g)
XI	64.11 ± 0.75c	0.84 ± 0.07d	1.99 ± 0.17ab	0.31 ± 0.03ab	3.68 ± 0.48abc	29.07 ± 1.30d
GH	68.47 ± 0.67b	0.81 ± 0.03d	2.19 ± 1.06a	0.31 ± 0.02ab	3.45 ± 0.37abcd	24.78 ± 0.78ef
CB	59.87 ± 0.10e	0.83 ± 0.02d	1.74 ± 0.06abc	0.30 ± 0.01ab	2.75 ± 0.02cde	34.53 ± 0.08b
SV	55.40 ± 0.93f	0.79 ± 0.03d	1.45 ± 0.27bc	0.28 ± 0.00ab	2.28 ± 0.06e	39.73 ± 1.34a
HO	62.39 ± 0.19d	1.20 ± 0.01a	1.78 ± 0.07abc	0.36 ± 0.04a	3.03 ± 0.20bcde	31.33 ± 0.07c
LB	64.70 ± 1.85c	0.84 ± 0.02d	1.53 ± 0.19abc	0.27 ± 0.02ab	2.94 ± 0.04bcde	29.67 ± 1.53cd
TQ	53.25 ± 0.70g	1.05 ± 0.12ab	1.24 ± 0.24c	0.33 ± 0.10ab	3.93 ± 0.40ab	40.29 ± 0.14a
NG	69.74 ± 0.35b	1.02 ± 0.10bc	1.76 ± 0.08abc	0.24 ± 0.04b	4.09 ± 1.12a	23.14 ± 0.80fg
MA	68.32 ± 1.38b	0.83 ± 0.02d	1.45 ± 0.1abc	0.25 ± 0.01b	3.63 ± 0.44abcd	25.47 ± 1.07e
LU	72.39 ± 0.25a	0.87 ± 0.12cd	1.78 ± 0.26abc	0.30 ± 0.02ab	3.13 ± 0.61abcde	21.52 ± 0.66g
TI	59.71 ± 0.08e	0.96 ± 0.19bcd	1.36 ± 0.31bc	0.32 ± 0.09ab	2.95 ± 0.35bcde	34.73 ± 0.27b
CA	58.60 ± 0.36e	0.86 ± 0.04cd	1.98 ± 0.06abc	0.29 ± 0.04ab	3.35 ± 0.65abcd	34.90 ± 0.22b
ST	58.75 ± 0.47e	0.90 ± 0.08bcd	1.89 ± 0.31abc	0.33 ± 0.01ab	2.64 ± 0.03de	35.49 ± 0.12b

Note: Different letter groups in the same indicator and analysis object showed statistically significant differences (p < 0.05).

On the other hand, the protein contents of Sap Vang (1.45 g/100g), Ta Qua (1.24 g/100g), and Tieu (1.36 g/100g) were significantly lower compared to the other banana varieties (p < 0.05). However, the protein content of the remaining varieties, ranging from 1.24 to 2.19 g/100g, is still higher than that of four varieties – Fenjiao, Dajiao, Red banana, and Haigongjiao – grown in China (1.01 – 1.75 g/100g) (Tan et al., 2024). This also showed that the 13 Vietnamese banana varieties have better nutritional potential as protein plays many structural and functional roles in the human body (Ranjha et al., 2022).

Additionally, the results showed no statistically significant differences in fat content (p > 0.05). Specifically, the fat contents of the 13 banana varieties ranged from 0.24 to 0.36 g/100g. Overall, the fat content of the banana varieties in this study is similar to that of the Musa variety (0.33 g/100g) reported in a previous study (Ogunlade et al., 2019). Furthermore, the fiber content showed significant differences between the varieties in this study, with a range of 2.28 to 4.09 g/100g. This result is similar to the fiber content of Indian banana varieties (Musa AAA spp., Musa AAB spp., and Musa AB spp.) (2.99 – 3.95 g/100g) except for Ngu banana (0.17 g/100g) (Dhandapani et al., 2017), and higher than the fiber content of the Mchare banana variety (0.92 – 2.79 g/100g) (Dotto et al., 2019). Given current nutritional recommendations encouraging the consumption of low-fat, high-fiber foods for improving gut health and managing metabolic disorders (Fritsch et al., 2021; Lattimer & Haub, 2010), the banana cultivars analyzed in this study may be considered beneficial dietary options that align with these health trends.

The results also showed changes in the carbohydrate content of the banana varieties studied. The carbohydrate content ranged from 4.63 to 15.23 g/100g, which is lower than that of Musa spp. fruit ranging from 21.8 to 32 g/100g (Aurore et al., 2009). Ta Qua banana (40.29 g/100g) and Lun banana (21.52 g/100g) were found to have the highest and lowest carbohydrate contents, respectively, with significant statistical differences (p < 0.05). Moreover, some carbohydrates in bananas are believed to accelerate the calorie-burning process in the body due to the presence of short-chain fatty acids (Wan et al., 2020). Studies have also found that these fatty acids can improve the body's nutrient absorption, especially calcium (Jenkins et al., 1998).

3.5. Physicochemical properties of banana varieties

In addition to basic nutritional composition, physicochemical properties play a vital role in determining the sensory quality, palatability, and processing suitability of banana cultivars. Key parameters such as titratable acidity, pH, total soluble solids, and reducing sugar content influence taste, texture, fermentation potential, and product stability.

As shown in Table 4, significant differences (p < 0.05) were observed among the cultivars for all evaluated properties. These variations highlight the distinctive biochemical traits of each banana variety, underscoring the need for cultivar-specific strategies in product development and processing optimization. Indeed, the TA values of the 13 Vietnamese banana varieties tested showed statistical significance (p < 0.05). The TA values ranged from 0.14 to 0.72 g/100g, which tended to be lower than those of Cavendish bananas, which ranged from 1.475 to 2.335 g/100g (Yap et al., 2017). The low acidity level indicated that the citric acid content in the fruit flesh was low. TA values reflect the overall concentration of hydrogen ions (both dissociated and undissociated) and are more closely related to flavor than pH values. Similarly, the banana varieties showed significant differences in pH values of the fruit flesh, with values ranging from 4.05 to 5.89. The highest pH values were found in the Sap Trang (5.89), Cau (5.92), and Ta Qua (5.69) bananas, and although there were differences in pH values among these three varieties, the differences were not statistically significant (p > 0.05). The pH of food is determined by the amount of hydronium ions present in the food product (Anyasi et al., 2015). A report found that the lower pH in bananas is due to an increase in organic acids as the fruit ripens (Arvanitoyannis & Mavromatis, 2009). Therefore, pH values can be used as a criterion for assessing ripeness at banana harvest. Furthermore, many food quality criteria have been found to correlate more strongly with pH values than with acid concentrations (Anyasi et al., 2015).

Table 4.

Physicochemical properties of banana varieties

Banana varieties	Titratable acidity (g/100g)	pH	TSS (°Bx)	Reducing sugar (g/100g)
XI	0.22 ± 0.02fg	5.06 ± 0.28a	22.03 ± 0.10a	26.54 ± 0.55bc
GH	0.51 ± 0.04b	4.59 ± 0.06d	22.36 ± 0.17a	42.63 ± 1.33a
CB	0.21 ± 0.02fg	4.11 ± 0.02ef	22.01 ± 0.02a	15.94 ± 0.37ef
SV	0.18 ± 0.01gh	5.49 ± 0.08b	22.31 ± 0.13a	13.82 ± 0.29fg
HO	0.43 ± 0.04c	4.05 ± 0.08f	22.92 ± 0.09a	12.29 ± 0.03g
LB	0.39 ± 0.05cd	5.07 ± 0.08c	22.71 ± 0.03a	27.76 ± 2.94b
TQ	0.17 ± 0.01gh	5.69 ± 0.04ab	22.13 ± 0.74a	14.57 ± 0.09ef
NG	0.35 ± 0.01de	5.19 ± 0.08c	22.02 ± 0.23a	28.75 ± 0.51b
MA	0.25 ± 0.01f	4.59 ± 0.13d	22.34 ± 0.52a	16.41 ± 0.69e
LU	0.72 ± 0.04a	4.32 ± 0.16e	22.87 ± 0.42a	25.02 ± 1.74c
TI	0.33 ± 0.01e	4.77 ± 0.12d	22.97 ± 0.11a	21.81 ± 0.43d
CA	0.20 ± 0.02fg	5.92 ± 0.04c	22.13 ± 0.17a	14.07 ± 0.11fg
ST	0.14 ± 0.02h	5.89 ± 0.04c	22.01 ± 0.11a	28.35 ± 0.29b

Note: Different letter groups in the same indicator and analysis object show statistically significant differences (p < 0.05).

The results showed that the TSS values varied but were not significantly different among the banana varieties (p > 0.05), ranging from 22.00 to 22.97 °Bx. In principle, TSS represent the total soluble solids in the fruit, and high TSS is associated with high sucrose concentrations in the fruit flesh, along with dry matter and other factors. At the same time, the TSS parameter also affects the flavor of bananas (Obenland et al., 2010). Similarly, the results also showed that the different banana varieties had significantly different reducing sugar contents (p < 0.05). The reducing sugar content of the 13 Vietnamese banana varieties evaluated in this study ranged from 12.29 to 42.63 g/100g. Low reducing sugar content and high TSS are typical characteristics of ripe bananas due to increased conversion of starch to sugar, with reducing sugars serving as substrates for the banana’s respiration process (Watharkar et al., 2021).

3.6. Bioactive components of banana varieties

The protective effects of phenolic compounds in biological systems arise not only from their ability to scavenge free radicals and chelate metal ions, but also from their capacity to modulate endogenous antioxidant enzyme activities (Olszowy, 2019). Besides, the observed variation in the levels of phenolics, flavonoids, ascorbic acid, and carotenoids among banana cultivars highlights their differing antioxidant capacities and potential health-promoting properties.

The results showed that the content of bioactive compounds was significantly influenced by the different banana varieties. The total phenolic content of the 13 banana varieties ranged from 0.73 to 11.48 mgGAE/gDW (Figure 2). A previous study reported that the highest total phenolic content was found in the Ducasse fruit at 1.28 mgGAE/gDW, and the lowest was in the Red Dacca and Plantanin fruits, with values of 0.40 and 0.38 mgGAE/gDW, respectively (Bashmil et al., 2021). Overall, the total phenolic content of six banana varieties (Cavendish, Ladyfinger, Red Dacca, Plantanin, Monkey, Ducasse) grown in Australia was relatively lower compared to the 13 banana varieties studied in this research (0.73 to 11.48 mgGAE/gDW). Similarly, a previous report tested 11 different banana varieties and found that bananas contain many antioxidants, with Red bananas (∼1.30 mg/gDW) and Nendran bananas (∼3.60 mg/gDW) showing higher antioxidant activity compared to Grand Naine bananas (1.13 mg/gDW) (Kumar et al., 2023). These findings are in agreement with those observed in several banana cultivars collected in Vietnam. Additionally, ascorbic acid (which is abundant in fruits) is both one of the essential vitamins (which humans cannot synthesize due to a lack of the enzyme L-gulonolactone oxidase) and a strong antioxidant and anti-inflammatory agent (Gęgotek & Skrzydlewska, 2022; Mazurek & Pankiewicz, 2012). A correlation was observed between total phenolics and ascorbic acid. Changes in banana variety were associated with significant differences in ascorbic acid content (p < 0.05), with values ranging from 0.12 to 0.81 mg/100gDW. Among them, the Lun banana had the highest total phenolic content (11.48 mgGAE/gDW) and ascorbic acid content (0.80 mg/100gDW). In contrast, the Sap Trang banana had the lowest total phenolic content (0.73 mgGAE/gDW) and ascorbic acid content (0.19 mg/100gDW) compared to the other 12 banana varieties.

Figure 2.

Total bioactive compounds content of banana varieties.

Flavonoids represent a major class of phenolic compounds, widely recognized for their diverse biological activities. In oxidative systems, they act as effective free radical scavengers, inhibit lipid peroxidation, and regulate redox-sensitive signaling pathways (He et al., 2017). Moreover, flavonoids have been shown to modulate the activity of key antioxidant enzymes such as superoxide dismutase, catalase, and glutathione peroxidase, thereby enhancing the cellular defense mechanisms against oxidative stress (Ighodaro & Akinloye, 2018; Orabi & Abou-Hussein, 2019). Their presence in bananas not only contributes to antioxidant capacity but also potentially supports the prevention of oxidative damage-related disorders. Flavonoid compounds were found to be higher in the peel than in the flesh of the banana varieties, with Ducasse (0.03 mgQE/gDW) being common but relatively lower than the other banana varieties studied (Bashmil et al., 2021). The Ngu banana was found to have the highest flavonoid content (0.40 mgQE/gDW). Meanwhile, three banana varieties, Cau, Cha Bot, and Sap Trang, had the lowest flavonoid content, with values of 0.12 mgQE/gDW.

Additionally, the carotenoid content provides health benefits due to its unique physiological functions, such as its role as a provitamin and antioxidant, especially in removing singlet oxygen. The results showed significant differences in carotenoid content (p < 0.05), ranging from 8.63 to 60.25 mg/100gDW. Among them, the Hot banana (157.17 mg/100gDW) had the highest carotenoid content, followed by the Sap Vang banana (60.25 mg/100gDW) and the Sap Trang banana (53.62 mg/100gDW). In contrast, varieties such as Cau, Gia Huong, Laba, Ta Qua, Ngu, Mat, and Lun had the lowest carotenoid content, ranging from 8.63 to 12.14 mg/100gDW. However, the content in these varieties was still significantly higher than that of the Fe’i banana variety grown in Asupina, which has yellow/orange flesh and contains only 0.00706 mg/100gDW (Englberger et al., 2006). This demonstrates that carotenoid content is influenced by the color of the fruit flesh, with the highest content found in bananas with yellow/orange flesh.

In general, the significant changes in the content of bioactive compounds across different banana varieties are attributed to both internal and external factors, such as genetic characteristics, plant variety, soil composition, care conditions, and post-harvest treatments (Faller & Fialho, 2010). Similarly, other studies have also reported differences in the content of bioactive compounds, such as phenolics, flavonoids, and ascorbic acid, in three banana varieties (Musa sapientum, Musa acuminata, Musa acuminate) (Adedayo et al., 2016). Additionally, some studies suggest that carotenoid content varies by banana variety (Heng et al., 2017; Kargar et al., 2019).

3.7. Antioxidant activity of banana varieties

DPPH and ABTS free radicals are considered stable free radicals and are widely used in experiments to estimate the free radical scavenging activity of materials with antioxidant activity (Kedare & Singh, 2011; Rumpf et al., 2023). The DPPH and ABTS antioxidant activities are affected by different banana varieties (Figure 3).

Figure 3.

Antioxidant activity of banana varieties.

The results demonstrated that variations in banana variety led to significant differences in DPPH and ABTS antioxidant activities. The antioxidant activity of the 13 banana varieties ranged from 0.51 to 1.67 mgAA/gDW (DPPH) and 0.30 to 1.17 mgAA/gDW (ABTS) (Figure 3). A previous study investigated two banana varieties, M. sinensis and M. paradisiaca, and the results showed higher ABTS and DPPH antioxidant activities compared to the 13 banana varieties studied in this research (Gervásio & Batitucci, 2023). The study by Adedayo et al. (2016) also reported similar findings, highlighting that antioxidant activity in bananas varies significantly between different varieties. Specifically, it pointed out that bananas possess considerable antioxidant activity, but the level varies depending on the cultivar. Indeed, the Lun banana stood out with the highest DPPH (1.67 mgAA/gDW) and ABTS (1.17 mgAA/gDW) antioxidant activities compared to the other 12 varieties (0.51 – 1.67 mgAA/gDW for DPPH and 0.30 – 1.17 mgAA/gDW for ABTS). Overall, the 13 banana varieties studied in this research all exhibited notable antioxidant activity. This is consistent with findings from a previous study, which reported that bananas possess antioxidant potential (Shian et al., 2012). Meanwhile, the Sap Trang and Xiem bananas were the two varieties with the lowest DPPH and ABTS antioxidant activities, respectively, at 0.50 mgAA/gDW, 0.82 mgAA/gDW, 0.54 mgAA/gDW, and 0.30 mgAA/gDW. These results are lower than the antioxidant activity reported for certain banana cultivars (Mas and Awak bananas) previously harvested in Malaysia, which ranged from 19.39 to 22.57 mgTE/gDW (Sulaiman et al., 2011).

3.8. Hierarchical clustering of banana varieties

The hierarchical clustering diagram (Figure 4) helped group banana varieties based on 18 evaluated parameters, including color (L*, a*, b*), nutritional composition (moisture content, protein, lipid, ash, and fiber), physicochemical characteristics (total acid content, reducing sugars, pH, and total soluble solids), and biological characteristics (phenolics, flavonoids, vitamin C, carotenoids, ABTS, and DPPH antioxidant activities). The clustering analysis results showed the formation of three main clusters with smaller subgroups, indicating similarities among banana varieties (Figure 6). Cluster 1 included the banana varieties XI, SV, ST, TQ, NG, and CB. In this cluster, CB was not grouped with the others, indicating distinct differences between CB and the remaining varieties. Furthermore, XI, SV, and ST belonged to the larger size group compared to TQ and NG, with significant variations among them. In contrast, TQ and NG showed smaller changes, reflecting similarities in genetic characteristics or chemical composition. Moreover, the two varieties XI and SV formed the smallest subgroup at the bottom, suggesting they were the most similar to each other. In summary, these differences indicated rapid variation among the banana varieties. Cluster 2 included the banana varieties CA, GH, TI, LU, MA, and LB. In this cluster, GH, TI, LU, and MA shared similar characteristics. Indeed, the fruit flesh color, structure, and appearance were comparable. Fewer similarities were found in the CA and LB groups. Similarly, GH, TI, and LU also shared similar characteristics, with GH and TI forming the smallest subgroup, while LU was in a larger but closer group. However, MA showed significant differences as it belonged to a larger group compared to GH, TI, and LU. One interesting observation was the overlap of variations in the clustering from bottom to top, indicating a high degree of similarity among these banana varieties. Finally, the HO variety stood out with substantial differences, showing that HO varied significantly from the other 12 banana varieties.

Figure 4.

The hierarchical clustering dendrogram groups the banana varieties based on 18 evaluated parameters, including color (L*, a*, b*), nutritional composition (moisture content, protein, lipid, ash, and fiber), physicochemical characteristics (total acid content, reducing sugars, pH, and total soluble solids), and biological characteristics (polyphenol content, flavonoids, vitamin C, carotenoids, ABTS, and DPPH antioxidant activities).

Figure 6.

Correlation matrix of physicochemical and bioactive variables visualized using a color-coded correlogram.

Overall, observations indicated that while the banana varieties differed, they shared many similarities. Specifically, GH and TI were grouped in the smallest subgroup, showing a high level of similarity between them. Additionally, similarities between varieties were evident when grouped into clusters; Cluster 1 (XI, SV, ST, TQ, NG, and CB) and Cluster 2 (GH, TI, LU, MA, and LB). These results demonstrated the intra-group similarities. These findings provided a comprehensive overview of the relationships among the different banana varieties.

Principal Component Analysis (PCA) illustrated the distribution of samples and the contribution of different variables to the three principal components of the Vietnamese banana varieties. PCA is a data reduction method that helps optimize the retention of important information. This technique transforms data from the original high-dimensional space into a new space with fewer dimensions (the principal components), thus facilitating the observation, analysis, and clarification of relationships between various indices, including color (L*, a*, b*), nutritional composition (moisture content, protein, lipid, ash, and fiber), physicochemical properties (total acid content, reducing sugars, pH, and total soluble solids), and biological properties (phenolics, flavonoids, vitamin C, carotenoids, ABTS, and DPPH antioxidant activities). Additionally, PCA highlights the differences among the indices of color components (L*, a*, b*), nutritional components (moisture, protein, lipid, ash, and fiber), physicochemical properties (Titrated acidity, reducing sugars, pH, and total soluble solids), and biological properties (phenolics, flavonoids, vitamin C, carotenoids, ABTS, and DPPH antioxidant activities). In this study, the first seven principal components (PC1 – PC7) explained 23.80%, 20.30%, 12.70%, 10.00%, 5.90%, 5.20%, and 4.5% of the total variance in the data, respectively, with a total of 82.40% variance explained (Supplementary Figure 2). This indicated that, with just these seven principal components, most of the variation in moisture content, phenolics, flavonoids, vitamin C, antioxidant activity (DPPH and ABTS), and color indices (L*, a*, b*) could be described.

Based on the loading vectors across the first seven principal components (PC1–PC7), a comprehensive interpretation of the multidimensional variation among banana samples was established (Supplementary Table 1 and 2). PC1 was predominantly influenced by DPPH antioxidant activity, moisture content, titrated acidity, ABTS activity, vitamin C, polyphenols, and total flavonoids. These variables captured the major variance associated with antioxidant capacity, internal moisture, and bioactive compound concentration. Notably, antioxidant parameters showed positive correlations with moisture and fiber content, while color indices such as redness of peel (a* - P) and carotenoids exhibited negative loadings, suggesting a trade-off between antioxidant richness and visual pigmentation. PC2 was largely driven by physical and structural attributes, including fruit weight, length, gumminess, chewiness, hardness, and lightness of both peel and flesh. This component reflected the morphological and textural characteristics of the fruit, highlighting distinctions in size, firmness, and external appearance. PC3 exhibited strong contributions from ash content, vitamin C, carbohydrate, and yellowness of peel (b* - P), indicating its association with mineral composition, nutritional content, and coloration. Meanwhile, lipid and protein showed negative loadings, pointing to a contrast between macronutrient density and ash-associated properties. PC4 was characterized by high loadings from cohesiveness, resilience, flavonoids, fiber, and redness of peel, indicating an interrelationship between mechanical resistance and bioactive content. This component emphasized the structural quality of the banana flesh in relation to antioxidant parameters. PC5 was mainly associated with yellowness of flesh (b* - F), pH, and firmness, representing a combination of visual ripeness, acidity, and textural density. The direction of loadings in this component suggested differences in ripening stages and internal consistency. PC6 displayed opposing contributions from carbohydrate, polyphenols, and lightness of flesh (positive) versus protein and reducing sugars (negative), reflecting a contrast between energy-storage compounds and structural protein content. Finally, PC7 was defined by lipid (positive), fiber (negative), cohesiveness, and resilience, highlighting variation in fat-related texture and elastic properties of the banana samples. Overall, the first seven principal components accounted for a substantial proportion of the total variance and offered deeper insights into the complex interplay between biochemical composition, antioxidant potential, texture, and visual characteristics among banana varieties.

The correlation analysis results of PCA and Pearson’s correlation showed that components including moisture content, DPPH, and titratable acidity had a strong positive correlation (p < 0.001) (Figure 5, Figure 6). According to a study by (Mathias et al., 2024), bananas contain certain acids that affected their free radical scavenging capacity. Antioxidant activity changes depending on the banana variety. Therefore, bananas with lower titratable acidity tended to have reduced antioxidant activity. Additionally, antioxidant activity was positively influenced by moisture content. A previous study showed that as bananas ripened, antioxidant capacity increased (Sulaiman et al., 2011). Indeed, as bananas ripened, starch in green bananas gradually converted into sugars and water, increasing the fruit’s moisture content. These results indicated that high moisture environments help preserve and optimize the antioxidant capacity in bananas. However, to ensure quality, if moisture content exceeds the optimal level, excessive oxidation or microbial growth could occur.

Figure 5.

The PCA biplots show variable contributions as vectors and sample clustering with ellipses. The selected components explain over 80% of total variance, highlighting clear group separation based on physicochemical and bioactive traits.

Flavonoids contributed significantly to the antioxidant potential of bananas. From the correlation analysis, a positive correlation was found between flavonoids, phenolics, ABTS, and vitamin C. These components all demonstrated strong antioxidant capacity (Ejaz et al., 2023). Many previous studies indicated that flavonoids are a subclasss of total phenolics (Arora et al., 2019). A previous study on several types of leaves and seeds also demonstrated a strong positive correlation between flavonoids, phenolics, and ABTS antioxidant activity (R² = 0.97 to 0.99) (Kiani et al., 2023). This correlation of these components was also similarly observed in a study on the leaves of Australian Myrtles (Ali et al., 2024). Overall, this combination may create a synergistic effect, enhancing overall antioxidant potential.

Moreover, the correlation between carotenoids, pH, a*, and TSS also showed a strong correlation. This relationship reflected specific color characteristics. The carotenoid content in bananas determines the yellow or orange level of the fruit flesh. The a* value also contributes to the color of the fruit’s peel and flesh, determining the redness of the banana, which explains the increase in a* as the banana’s color gradually turns red. On the other hand, the correlation analysis, showed that color values had a negative correlation with nutritional values.

Moreover, Figure 5 revealed a substantial similarity among the 13 evaluated banana groups, represented by overlapping ellipses on the PCA diagram. This indicated that these banana cultivars shared several similar characteristics and that none of them was completely distinct from the others. However, each cultivar still exhibited its own unique chemical, biological, and physicochemical traits. Among them, cultivars CA and NG were identified as the most distinct, while CB and HO represented the cultivars that reflected the overall characteristics of the other 11 cultivars, in addition to their own distinctive features. These findings were further supported by the clustering pattern revealed in the hierarchical clustering dendrogram (Figure 4).

Besides, the compositional differences observed among the 13 banana cultivars may result from a combination of genetic, environmental, geographical, and postharvest factors. Genetic variation affects the biosynthesis of bioactive compounds such as phenolics and carotenoids, with cultivars from different genomic groups exhibiting distinct metabolic profiles. Environmental conditions (e.g., soil, altitude, and climate) influence nutrient uptake and stress responses, while geographical factors shape agro-ecological interactions. These combined influences explain the diversity seen in PCA and clustering results and should be considered in future research to separate genetic effects from external variables.

4. Conclusion

The components in bananas contained compounds that are important for the human body. The findings from this study highlighted significant differences in nutritional composition, physicochemical properties, and antioxidant activity between 13 Vietnamese banana varieties. These findings suggested that certain banana cultivars grown in Vietnam may be comparable to, or even outperform, some internationally cultivated varieties. Indeed, the nutritional components – particularly the bioactive compounds – of several Vietnamese cultivars were found to be considerably higher than those reported for Ducasse bananas (Malaysia) and Ladyfinger bananas (Australia)… Varieties such as Ta Qua and Sap Vang showed notable differences in fruit size, nutritional content, and levels of bioactive compounds, emphasizing the potential of certain varieties to become superior sources of nutrition. The study reaffirmed the value of Vietnamese bananas in supplementing daily nutrients and provided a basis for selecting specific varieties with higher health benefits for functional food applications. Further research could explore environmental factors affecting these traits, offering deeper insights into optimizing banana production for nutritional and functional purposes.

CRediT authorship contribution statement

Ngoc Duc Vu: Writing – review & editing, Writing – original draft, Visualization, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Nguyen Trinh Thi Nhu Hang: Software, Formal analysis. Kieu Dang Quynh Mi: Software, Formal analysis. Nguyen Huu Thuan Anh: Resources, Project administration. Binh An Pham: Software, Project administration, Formal analysis.

Funding

This research is funded by Nguyen Tat Thanh University, Ho Chi Minh City, Vietnam under grant number 2024.01.131/HĐ-KHCN

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material 1

The external appearance of 13 banana varieties in Vietnam.

Supplementary material 2

Scree plot showing the eigenvalues of principal components used to determine the number of meaningful components.

Supplementary material 3

Mean and standard deviation values are listed alongside the loading coefficients of 30 variables across the first seven principal components, indicating each variable’s contribution to the PCA model.

Supplementary material 4

Scores represent the projection of each observation in the multivariate space defined by the first seven principal components, reflecting underlying patterns and group structures.

Data availability

All the data are available within this manuscript