Abstract

Irvingia gabonensis and I. wombolu are two Irvingiaceae species with economic significance, because of their valuable kernels for West and Central Africans to thicken sauces. They are called sweet/edible – and bitter/inedible – fruited African bush mango trees (ABMTs), respectively, because of the taste of their mango – like fruit. The striking difference in fruit pulp taste clearly discriminates between Irvingia gabonensis and I. wombolu trees, which are morphologically very similar. Experts profoundly disagree on the taxonomic significance of differences among bush mango trees to allow classification of Irvingia gabonensis and I. wombolu as clearly different species. Therefore, the random species or variety consideration in researches indicates difficulty to define taxonomic units, which is vital for effective conservation and use of ABMT germplasm, because the source of tree – to tree variations is uncertain. This paper reviews ethnobotanical and biochemical data on ABMTs in order to evaluate their contribution to the biosystematics of I. gabonensis and I. wombolu. Thus, information on local uses (food and medicinal applications) and biochemical data of the kernel, pulp, leaves and bark are compared, and their implications for the distinction between I. gabonensis and I. wombolu are discussed. We evidenced a lack of comparative data related to ethnobotanical and many phytochemical aspects, mostly on the bark, leaves and roots. We hypothesised obvious misidentifications causing conflicting patterns, but no evidence of any phytochemical dissimilarity was found between I. gabonensis and I. wombolu. Kernel macronutrient composition, diversity, types and abundance of fatty acids in the kernel oil and the functional and therapeutic properties of this oil do not allow to discriminate between both species. Moreover, there is no correlation among patterns of lipophilic components profile, composition in fatty acids in the oil, geographical distribution of ABMTs. We interpreted these random patterns within ABMTs as tree – to – tree variations and this explains the vast co-occurrence of both taxa. However, consistent differences in fruit pulp taste across contrasting ecological conditions imply significant phytochemical differentiation between I. gabonensis and I. wombolu. This needs to be thoroughly re-evaluated and correlation with genetic data will clarify the taxonomic issue.

1. Introduction

African bush mango trees (ABMTs) widely occur in natural forests of West and Central Africa. They are very important fruit tree species that are intensively cultivated by local farmers due to the economic significance of the kernels they produce ¹. These slimy kernels are used by millions of Africans as sauce – thickening ingredient ². The kernels also produce edible oils usable in human nutrition and cosmetics ^{3, 4}. The fruit pulp tastes sweet and is eaten, but extremely bitter and unpalatable fruits pulp also exist ⁵. However, kernels with better organoleptic properties (highly slimy sauce with intensive aroma) come from the extremely bitter ^{6, 7}.

Species delimitation is central to biodiversity classification and conservation, while species definition concepts are numerous and controversial ^{8, 9}. But, genetic isolation, which is a core idea in biological species definition, plays significant role in speciation processes and thus, is important to all species concepts ¹⁰. A simplified species idea in taxonomy defines species as “a group of kinds of interbreeding natural populations that are reproductively isolated from other such groups” ^{9, 11}. Here, the edibility of the bush mango pulp defines two reproductively isolated kinds of ABMTs that are, however, sympatric and botanically very similar ^{12, 13}. Applicability of morphological species ¹⁴ eases taxa recognition and facilitates domestication and efficient use of biodiversity ¹⁵. Difficulties to separate sweet – fruited ABMTs from the bitter ones, based on herbarium vouchers and living materials, is source of species problem within ABMTs. Sweet – and bitter – fruited ABMTs were considered ^{16, 17}as two varieties within a unique Irvingia gabonensis species: Irvingia gabonensis (Aubry-LeComte ex O’Rorke) Baill. var. gabonensis and I. gabonensis var. excelsa Okafor, respectively. However, ⁵ revised the Irvingiaceae family and raised the sweet – and bitter – fruited ABMTs at species level, based on the correlation between the striking differences in fruit pulp taste with those in the fruits, seeds, and leaves: Irvingia gabonensis and I. wombolu Vermoesen, respectively. This opinion was later confirmed by significant genetic and phenological dissimilarities between these two kinds of trees ^{13, 18}. However, the species consideration by ¹⁸ is still not totally acknowledged, since fruit taste and morphological diversity are often considered as random tree – to – tree variations within ABMTs ¹. The species’ problem within ABMTs makes uncertain the definition of biological units for conservation and increases uncertainty in management and domestication programs ¹⁹. For example, extinction of I. wombolu is taking place in the Volta Forest region (Ghana – Togo), because this kind of ABMT is considered by conservationists as wild I. gabonensis trees ^{13, 20}. Consequently, it is difficult to: (i) develop efficient market led – domestication and cultivation systems; (ii) develop ABMTs – based food and pharmaceutical products at a wider scale; and (iii) suggest policies for the efficient conservation of ABMTs.

Biosystematics combines a wide range of biological criteria (ecology, morphology, phenology, genetics, phylogenetics, eco-physiology, etc.) to capture patterns of variations and classify organisms, based on the correlation between differences ^{21, 22}. Comparative ethnobotanical data may provide hypotheses regarding differences in biochemical characteristics, to support plant taxonomic decision ²³, because plant biochemistry, genetics and ecological circumstances are inextricably linked ^{24, 25} (Pacheco-Hernández et al., 2021). Therefore, ²⁶ combined phytochemical and genetic data to clarify the delimitation of a number of bryophyte species and ²⁷ identified significant genetic and morphological differences among folks classified potato materials.

This review evaluates the contribution of available ethnobotanical and biochemical data to the biosystematics of I. gabonensis and I. wombolu. Thus, information on bitter – and sweet – fruited trees was compared, to see how both differ in food and medicinal applications as revealed by ethnobotanical data, on one hand. On the other hand, the two types of ABMTs were compared for the biochemical properties of their kernel, fruit pulp, leaves and bark.

2. Methods

We performed a literature search in the databases of World Agroforestry Centre, African Journal Online (AJOL), Scopus, PubMed and Web of Science. Only papers dealing with ethnobotany, macronutrients, fatty acids, phytochemicals, or functional properties of I. gabonensis and / or I. wombolu were analysed.

We recorded all types of uses with organs and types of ABMTs (I. gabonensis and / or I. wombolu) involved.

A database was built with different provenances (Congolian, upper Guinean and lower Guinean forest zones samples) of I. gabonensis and I. wombolu trees, for which data on kernel macronutrients content and fatty acids diversity plus concentration existed. Macronutrients included: proteins, lipids, minerals, ash, carbohydrates and moisture. We recorded fifteen identified fatty acids in this database: capric acid (C10:0), lauric acid (C12:0), myristic acid (C14:0), myristolic acid (C14:1), palmitic acid (C16:0), palmitolic acid (C16:1), stearic acid (C18:0), oleic acids (C18:1 and C18:1n9), linoleic acids (C18:2 and C18:2n-6), linolenic acid (C18:3), α-linolenic acid (C18:3n-3), arachidic acid (C20:0), docosanoic acid (C22:0) and erucic acid (C22:1). These data were supplemented with secondary plant metabolites contents.

Kernel oil physicochemical properties were also recorded: total saturated fatty acid, mono and polyunsaturated fatty acids contents, the oil saponification value (SV: mg KOH/g), its iodine value (IV: g I₂/100 g), peroxide value (meq O₂/kg) and acid value (mg KOH/g). Moreover, kernel oil physicochemical, functional and therapeutic properties were recorded as well as pulp, leaves, bark and root biochemical properties.

The two taxa were compared based on the percentage of the types of uses for which: (i) the two kinds ABMTs are involved, (ii) either I. gabonensis or I. wombolu was cited, and (iii) it was difficult to confirm the kind of ABMT that was studied, respectively. Misidentification was concluded based on three factors: material description (fruit pulp taste), geographical location and period of material collection ^{5, 13, 18}.

Using the package candisc ²⁸ of the R statistical software version 3.3. ²⁹, a step-by-step canonical discriminant analysis (CDA) was applied on oil plus fatty acid data, to determine parameters which discriminate the two kinds of Irvingia trees. A principal component analysis (PCA) was performed on the discriminant factors in order to promote visibility of the CDA results. Principal component scores from the first two axes of the PCA were plotted in two-dimensional space, in order to promote visibility. Significance of discriminate factors was evaluated, by means of one sample t test. Here, we tested hypothesis that the mean single value of each one of the totally independent factors for I. wombolu, is taken from I. gabonensis population ³⁰. This test was applied because of the limited sample size for I. wombolu. For this same reason, the one sample t test was also applied to total saturated, mono-saturated and poly-saturated fatty acids concentration in order to evaluate whether the value for I. wombolu could supposedly be considered as sampled from an I. gabonensis population.

Third, ³¹ established that the ratio C14:0 / C12:0 is always higher and lower than 1 for I. gabonensis and I. wombolu, respectively. In order to test this hypothesis at a broader geographical scale, a binominal logistic regression was applied on this ratio calculated on the fatty acid database, to determine the probability of any kernel sample to be collected from I. wombolu while the ratio is higher than 1 ³².

Kernel oil physicochemical, functional and therapeutic properties and pulp, leaves, bark and root biochemical properties were too limited, so that only descriptive analyses were performed.

3. Results

In total, forty different uses engaging various parts of ABMTs were recorded (Table 1). Irvingia gabonensis and I. wombolu were co-cited in six indigenous uses related to their kernels, cooked in identical food technological procedures ^{5, 20}, and their woody materials (trunk, branches and shells) valued in rural buildings, domestic energy, traditional fishing system, cultural activities and as chewing-stick (Table 1). Therefore, 15% of uses were similarity, or either identic between I. gabonensis and I. wombolu.

The fruit pulp of I. gabonensis is sweet and consumed as food and medicament, while that of I. wombolu is extremely bitter and was not cited for such utilizations ³³. Sour and inedible fruit pulp taste was also mentioned for I. wombolu ³⁴. These two utilizations were those that clearly differentiated both taxa. Most importantly, there was no specific indigenous use for I. wombolu (Table 1).

In 17.5% of the indigenous uses, the kind of ABMT to which the use was assigned could not be clearly identified, because the authors addressed Irvingia spp or ABMTs, without indicating whether it was the sweet – or bitter – fruited trees that were investigated. Moreover, the majority (62.5%) of indigenous uses recorded was mentioned for I. gabonensis, while it is not clearly demonstrated that: (i) I. wombolu possesses no such properties, (ii) and it was not I. wombolu which was not investigated, since material identification. Therefore, 80% of ethnobotanical uncertainty exists between I. gabonensis and I. wombolu. This uncertainty is related to therapeutic properties and is due to the lack of consistent material identification and comparative studies and ⁴ doubted whether there is any significant biochemical dissimilarity between both kinds of ABMTs.

It was most likely that specific medicinal properties may have been ascribed to the wrong material, leading to wrong biological hypotheses tested in laboratory analyses.

Macronutrient and fatty acid data were available for 42 trees: 34 and 8 I. gabonensis and I wombolu trees, respectively. Their kernel water content overlap: 2.1-12.8 and 6.7-11.9 for I. gabonensis and I. wombolu, respectively ^{35, 36, 37, 38}. Crude fat content in the kernels also overlaps: 37.5-76% and 38.4-69.7% for I. gabonensis and for I. wombolu, respectively, as evidenced by ^{2, 39, 40}. Similarly, variation in crude protein content in the kernel is identical for I. gabonensis and I. wombolu: 6.9-9.2% and 7.4-9.7%, respectively [38.40,41,42]. This similarity in the kernels’ crude fat (71.9 and 72.01 for I. gabonensis and I. wombolu, respectively) and protein content (8.65 and 8.66 for I. gabonensis and I. wombolu, respectively) is indeed clearly evidenced in a recent comparative study ⁴³. The overlap between I. gabonensis and I. wombolu is also observed for ash (1.5–11.1% against 2.5–10.33% for I. gabonensis and I. wombolu, respectively ^{37, 38, 42} and kernel carbohydrate content (7.4-39% and 26-41% for of I. gabonensis and I. wombolu, respectively: ^{35, 38, 44, 45}). Therefore, the clear separation between these two Irvingia species, based on macronutrients data ⁴⁶ was not demonstrated.

The kernels of both taxa mainly contain three types of saturated fatty acids: lauric, myristic and palmitic acids ^{31, 47, 48}. The CDA showed that all the variation was captured by the first canonical axis (99.5%). The canonical correlations with this axis (Table 2) are significant (df = 8, approximated F-statistics = 78.75, p-value = 0.002). The most discriminating parameters of the two species were C10:0, C12:0, C14:0 and C16:0, I. gabonensis having lower values of C10:0 and C12, but higher values of C14:0 and C16:0 while the opposite trend was observed for I. wombolu).

The PCA performed on these four discriminating fatty acids data indicated that the first two axes together account for 90.9% of the total variation (Table 3). The first axis (78.68%) was positively correlated with capric and lauric acids and negatively correlated with myristic and palmitic acids, while the second (12.22%) explains no consistent variation, compared to the first axis. Thus, two main groups of trees can be distinguished (Figure 1): (i) the first group (G1) most I. wombolu trees plus the two Nigeria I. gabonensis samples, with higher concentration of capric and lauric acids in their fruit kernels and lower concentrations of myristic and palmitic acids, and (ii) the second group (G2) made of most I. gabonensis kernels plus two I. wombolu samples, with opposite trends, as refer to G1. Indeed, Leakey ⁴⁹ indicated that concentration of those three major fatty acids (lauric, myristic and palmitic acids) is only provenance – dependent. Here, the PCA shows no such pattern. The groups are much more random than provenance-dependant.

The t test indicated that total oil content (t = 0.98; P = 0.33) and stearic acid concentration (t = 1; P = 0.22) in the kernels did not distinguish between I. gabonensis and I. wombolu. However, capric acid (C10:0; t = -14.36, P<0.001), lauric acid (C12:0; t = -11.28, P<0.001), myristic acid (C14:0; t = 3.19, P=0.004) and palmitic acid (C16:0; t = 8.36, P<0.001) significantly distinguished I. gabonensis from I. wombolu. Capric and lauric acids were significantly more concentrated in I. wombolu kernels, which contains less myristic and palmitic acids (Figure 1). This relationship between Capric, lauric myristic and palmitic acids is consistent throughout the Upper and Lower Guinean and Congolian forest regions ^{50, 51}.

Concentrations of saturated fatty acids in the kernel are similar for both taxa: 86.6–98.7% and 95.6–98%, respectively. These high concentrations of saturated fatty acids make of bad quality for cooking ⁵².

The logistic regression performed on R=C14:0 / C12:0 confirms that this ratio is linked to differentiation between I. gabonensis from I. wombolu (Type 3 effect: Chi²=16.99, P<=0.005). However, the power of this ratio to consistently separate both taxa is weak (Wald Stat = 56*10^-5, P=0.99). Reclassifying studied materials (the R ratio is lower than 1 for some I. gabonensis kernel samples, from the Upper and Lower Guinean Forest zones ^{48, 53, 54}, while it is higher than one for I. wombolu in the Congolian zone ⁵⁰, investigated by ³¹. Most importantly, in the same ecological conditions of Enugu State in West Nigeria: R > 1 and R < 1 were simultaneously found within I. gabonensis, depending on the village ⁴⁰ in the same ecological zone. Therefore, the variation of index is not linked to differences in the bush mango pulp taste, nor to any geographical distinction of provenances. This may better be considered as a simple tree-to-tree variation within ABMTs ².

Significant differences in secondary plant metabolites content, such as phytosterols may exist between the kernel oils of I. gabonensis and I. wombolu ³¹. Thus, beta sitosterol plus stigmasterol are abundant in the unsaponifiable fraction of I. gabonensis’ kernel oil, while it is the abundance of isobutyl phthalate plus stigmasta-3,5-dien-7-one that characterizes I. wombolu. These differences in secondary metabolites synthesis in oils are genetically-governed, and determine ecological fitness, growth system and survival of plant ⁵⁵. However, replication is needed to eliminate ecological and domestication effects for consistent conclusion.

The kernels of all provenances of I. gabonensis display linoleic acid contents close to 1%, while that of I. wombolu show significantly lower values of 0-0.4% ^{31, 50}. The difference with regard to the content of unsaturated fatty acids implies differences regarding ecological adaptation, and in particular growth under lower temperatures ⁵⁶. Unfortunately, results in West African samples contradict this conclusion. Therefore, the lack of differences in unsaturated fatty acid concentration explains the vast occurrence of I. gabonensis and I. wombolu.

Important variation exists, regarding the physicochemical properties of the kernel oil (Table 5.1), for both taxa. Acid value overlaps and ranges between 2.6-25.18 MgKOH/g for I. gabonensis and I. wombolu ^{57, 58, 59}. Similar variation was observed for peroxide and saponification values, refractive index, specific gravity, melting and smoke points ^{47, 48, 60}. The peroxide value in the kernels of I. gabonensis and I. wombolu overlap and range between 0.81 to 3.3 mg KOH/g, with no obvious difference between the two taxa. A similar situation is observed for the smoke point of 78–215 for both taxa. The refractive index and the specific gravity at 40°C range between 1.4 to 1.5 and 0.8 to 0.9, respectively for both taxa. In contrast significant differences in variation of iodine value were found: 2.54–3.5 and 4.2–32.43 for I. wombolu and I. gabonensis, respectively.

Few studies addressed the mineral composition of only I. gabonensis kernels and there are not comparative data with I. wombolu.

The comparative data on of the in vitro-produced callus of I. gabonensis and I. wombolu demonstrated that callus was formed after three weeks for both taxa. Moreover, proteins, carbohydrates, lipids, moisture, fibres, ash, iron, zinc, bore, cupper, potassium, calcium, manganese, phosphorus and nitrogen contents in the callus present no significant difference between I. gabonensis and I. wombolu ^{61, 62}.

Attempts to bind tablets using I. gabonensis defatted kernels demonstrated high brittleness plus friability values, with low disintegration and dissolution of the tablets. In contrast, gums from kernels of I. wombolu were successfully used to bind tramadol with short disintegration time ⁶³. Moreover, significantly higher thickening capacity was found for the defatted kernel of I. wombolu: 91% against 64% for I. gabonensis ⁴³, while evaluation of viscosity and drawability of the defatted kernels from I. gabonensis, contrastingly shows on the one hand, that the relationship between these two characters is weak, due to important tree-to-tree variation (3- to 5-fold) within this taxon ². On the other hand, this study also demonstrated that protein and fat contents did not have any effect on the thickening properties of I. gabonensis kernels. Unfortunately, this study did not include I. wombolu to allow direct comparison.

The consumption of kernel-based diets by diabetics (unknown doses and frequencies) may decrease blood glucose and increase activity of Na⁺, K⁺-ATPase, Mg²⁺-ATPase and Ca²⁺-ATPase ⁶⁴.

Precisely, 4g/day of I. gabonensis kernel during 4 weeks in type II diabetics and streptozotocin-induced diabetic treatment may significantly decrease plasma lipids, blood glucose, total cholesterol, triglycerides, HDl cholesterol, LVL + VLDL cholesterol and lower erythrocytes ATPases level with reduction of body weight and stabilization of activities of glycolytic enzymes pyruvate kinase, lactate dehydrogenase and glucose-6-phosphatase ^{64, 65, 66, 67}. Moreover, 3.15g of I. gabonensis kernel extracts administered three times a day during 4 weeks significantly decreased body weight and systolic blood pressure with many other positive specific effects ⁶⁸.

A standardized extract with 7% albumin from I. gabonensis kernels and extract a Cissus quadrangularis with 2.5% ketosteroids (Vitaceae) was applied to successfully control obesity ^{69, 70}. Here, the type of kernel (entire or defatted), type of extract (organic or aqueous) and doses remain unknown. It was also demonstrated that the kernel of I. gabonensis had anti-hypertensive, anti-lipidemic, anti-inflammatory, anti-nociceptive nephro-protective and hepatoprotective properties ⁷¹.

However, evidences of negative effects of the same diets exist: increased blood glucose and severe troubles in reproduction system: arrested spermatogenesis, degenerated germinal epithelium, absent interstitial cells, hardened basal cells and empty seminiferous tubules ⁷¹. Moreover, consumption of kernel extracts contrastingly increases of HDL-cholesterol ^{64, 65}. Unfortunately, there is no such data on I. wombolu.

The fruit pulp of I. gabonensis contains 0.8g/100g, 0.4g/100g, 1.09g/100g and 1.06g/100g of ash, crude fiber, crude proteins, and crude fat, respectively. It is also rich in calcium (262.3 mg/100g), total reduced sugar (459.7 mg/100ml), and ascorbic acid (67 mg/100ml) ^{72, 73}. Low concentration carotenoids (2.25mg/100g) was reported in the ripened fruit ⁷⁴. These surprisingly low concentrations of carotenoids call for thorough re-evaluation, because carotenoids determine fruit pulp colour ⁷⁵ which varies from yellowish to pink for I. gabonensis ⁷⁶. Again, Irvingia wombolu has not been evaluated for these characters.

Based on a screening, several phytochemical compounds were identified in the fruit pulp of I. gabonensis and I. wombolu ^{77, 78} (Table 5.2). In general, I. wombolu has higher alkaloids, saponins and tannins’ concentration according to ⁷⁸. However, lower concentrations of these phytochemical compounds or their total absence were contrastingly observed in the pulp of I. wombolu ⁷⁷. The pulp of I. wombolu contained 97.77±17.22 µg/g of total polyphenol compounds ⁷⁹ with phlobatannin, anthocyanin and quinones, which were specific to this taxon ⁷⁷. However, ⁷⁷ did not detect phlobatannins and quinones in the pulp I. wombolu, and ⁷⁸ missed anthocyanins that of I. gabonensis.

Here, the diversity and concentration of biochemical compounds of the fruit pulp present conflicting results and we could not conclude, any similarity / dissimilarity between I. gabonensis I. wombolu.

Concentrations of tannins, saponins, alkaloids, phenols, phlobatannins and anthraquinones were revealed through a screening of leaves of I. gabonensis ^{80, 81}. These studies also demonstrated that extracts of leaves of I. gabonensis significantly decreases gastrointestinal motility and increases gastric mucous secretion. Moreover, the extract inhibits gastric ulceration, decreases gastric acidity and shows significant anti-diarrheic properties ^{80, 82}. There is no research on the leaves from I. wombolu.

Terpenoids (3-friedelanone, betulinic, oleanicand hardwickiic acids) and ellagic acid derivatives (methyl gallate and 3-β- acetoxyursolic acids) were isolated from bark extracts of I. gabonensis ^{83, 84}. Alkaloids, tannins, phlobatannins, saponins, flavonoids, anthraquinones, phenols and cardiac glycosides were abundantly found in the bark of I. gabonensis ⁸⁵. These are important antioxidants, valuable in the treatment of cancerous tumours. ⁸⁶ demonstrated that the bark of I. gabonensis possesses analgesic properties. Antibacterial and antifungal activities of the bark of this taxon are used in traditional medicine as was successfully demonstrated by ⁸³. However, contrasting results were reported by ⁸¹ who observed no antimicrobial effects. ⁸⁴ presented in-vivo hepatoprotective and antioxidant activities of the bark extracts from I. gabonensis on the liver. Other studies demonstrated that significant loss of body weight, reduced subcutaneous fat and blood glucose plus HDL-cholesterol, increased liver enzymes activities were also noticed after the administration of the bark extracts ^{87, 88} Here again, we question these data since ⁸⁴ demonstrated hepatic degeneration. This calls for standardization (types of extracts, doses and frequency of consumption) to demonstrate the real effect of bioactive molecules against microorganisms and the limits of their use. A recent comparative analysis demonstrates significant differences between the two taxa, regarding phytochemical in the bark. However, this study is geographically limited and these differences are note easily perceptible. We postulate that these differences are not consistent, is in the case of macronutrients.

In West and Central Africa, the roots of I. gabonensis are traditionally used for a variety of purposes ^{13, 33}. But there is a lack of laboratory works to reveal the phytochemical substances that are responsible for these properties. High concentrations of tannins, saponins, alkaloids and anthraquinones are found in the root of I. gabonensis and it was demonstrated that 5mg/ml ethyl acetate extracts of it was highly active against various bacteria and fungus species ⁸¹. No data is available for roots of I. wombolu.

4. Discussion

This review shows a great diversity of ethnobotanical knowledge and a wide range of phytochemical properties of ABMTs. Thus, the bark, fruits, leaves, and roots of ABMTs contain substances with multiple anti-infective potential to control some complex multiple pathogens-induced and difficult to treat diseases ⁸⁹. However, three major concerns emerge from ethnobotanical and phytochemical data available on ABMTs.

First, a clear distinction between I. gabonensis and I. wombolu with comparative study are not often undertaken, because the two taxa are morphologically similar ^{40, 67, 90, 91, 92, 93, 94} and misidentification is suspected in many studies. For example, a sour taste appreciated in African diets ⁹⁵, was mentioned for the fruit pulp of I. wombolu ^{34, 96} and even used to distinguish I. wombolu from the sweet-tasting I. gabonensis ⁹⁷, while such a taste has never been perceived by consumers. Thus, botanical misidentification is a serious issue to be addressed in future research, since knowledge of the correct botanical and taxonomic position of plants is essential for a proper understanding of their ethnobotany, medical use, and conservation ⁹⁸. The pulp of bush mangoes always tastes either sweet or bitter. The unclear material identification takes roots in wrong ethnobotanical data collection strategies, because the endogenous biosystematics ²⁷ that leads to folk classification, on which hinges the better capture the existing variation.

Second, serious contradictions emerge from the biological data: increase and decrease of blood glucose, positive and negative impacts on human reproductive system, specificity of some phytochemicals which are totally missed in either I. gabonensis or in I. wombolu ^{77, 78}, absence of correlation among differences in iodine values, saturated fatty acids and secondary metabolites contents, etc. This absence of correlation supports the vast co-occurrence of I. gabonensis and I. wombolu in many ecological regions. Thus, ⁹⁹ to simply reject the conclusions of ^{68, 69, 70}. We suspected three primary causes that may lead to such a situation: (i) unreliable laboratory methods, (ii) absence of ecological variability in materials analysed, and (iii) completely random tree-to-tree variation within ABMTs. This variation of biochemical data can also be caused by different processing methods ^{51, 100, 101, 102}, nature of kernels (defatted or entire), and type of kernels used (from bitter or sweet fruits). Therefore, the contradictions are method-dependant and the ranges observed for all macronutrients and phytochemicals in this review are wider than differences claimed by ⁹⁹ to postulate any taxonomical significance between I. gabonensis and I. wombolu. Here again, the failure of ethnobotanical techniques to formulate correct biological hypotheses-based ¹⁰³ becomes danger in which all the causes of the contradictory biological data are rooted. ²⁷ indicates the importance of traditional biological knowledge to classify and geographically localize plant species. Consequently, ethnobotanical observations are also a taxonomic tool that provides background information for morphological, ecological, genetics and biochemical investigations.

Third, phytochemical analyses are often conducted on kernels acquired on local markets ^{38, 48, 104}. It is known that kernels from both taxa are often mixed for commercialization ¹⁰⁵ and no standard is available to accurately distinguish I. gabonensis from I. wombolu, based on the kernels alone. ⁷⁹ bought pulp of I. wombolu in local markets for phytochemical evaluation. There is no indication of extremely bitter fruits marketing, because kernels are always extracted at the collection place or at home. Thus, it is more likely that ⁷⁹ have studied I. gabonensis. Engaging genetically conflicting and morphologically similar materials, acquired on local markets, into biochemical analyses raises a third major concern, which is their nature (fresh or dried or pre-processed into flour or cake) in varying conservation conditions, affecting their biochemical composition ¹⁰⁶. Therefore, the side effects of the kernel-based sauce consumption ⁹⁹ and the conflicting biochemical data may have multiple causes that need to be clarified.

There is no clear distinction between I. gabonensis and I. wombolu based on their macronutrient’s composition. Thus, fruit pulp edibility might be an organoleptic variation that makes I. gabonensis of higher economic value and explain the vast data available on this taxon, and its priority status in traditional domestication and cultivation systems ¹⁰⁷.

The diversity, and concentration of fatty acids do not consistently separate I. gabonensis and I. wombolu. Likewise, the ration C₁₄ / C₁₂ fatty acids does not discriminate I. gabonensis from I. wombolu and other lipophilic components profiles need to be thoroughly reassessed. In addition to low botanical identity and vast co-occurrence of I. gabonensis and I. wombolu similar macronutrient and fatty acid characteristics evidence no taxonomic difference between these two taxa.

There is no geographical pattern within fatty acid data, demonstrating that differences in their biosynthesis is not environment mediated. Therefore, the lack of significant fatty acid difference between I. gabonensis and I. wombolu remains either laboratory technique-dependent or synonym taxonomic similarity. Differences of lauric, capric, palmitic and myristic acids concentration should be reassessed on a broader geographical scale for any consistent taxonomic conclusion.

Here, limited phytochemicals data (alkaloids, saponins and tannins) are conflicting, and lack of consistency, while they help clarify taxonomical decision in plant species ¹⁰⁸.

Total absence of sweetness plus hight and low bitte plus Fruit pulp and bitterness are specific to I. gabonensis and I. wombolu, respectively, regardless the ecological conditions of their growth. This character represents the main difference between the two taxa (Harris 1996), which may be ecologically different. Fruit taste is a genetically coded character ^{109, 110} and could be used to define subspecies / varieties / cultivars, but the environmental impact on its expression could be revealed significant ¹¹¹.

5. Conclusion

This review shows the extensive ethnobotanical and phytochemical data available for I. gabonensis and I. wombolu. It also demonstrates, the difficulty to correlate ethnobotanical data and macronutrient plus fatty acids and phytochemical properties to fruit taste, in order conclude any taxonomic distinction between I. gabonensis and I. wombolu. Moreover, functional and therapeutic properties of the kernel oils, which are plant biochemical expressions, do not discriminate both taxa. The random macronutrient and biochemical characteristics within ABMTs may imply genetic similarity between I. gabonensis and I. wombolu.

The overlap of phytochemical properties between closely related species is not unusual and given the extent of tree-to-tree variation within Irvingia gabonensis, it is possible that variation in taste, biochemistry and morphology may not be taxonomically significant. However, this should not obscure a range-wide DNA analysis of the tree-to-tree variations within ABMTs, regarding pulp taste and many other morphological and phytochemical traits.

Acknowledgements

This review is a part of the project: “Taxonomic revision of bitter and sweet African bush mango trees and priority setting for domestication in West Africa” founded by the Alexander von Humboldt Foundation (Scholarship to Vihotogbé R., first author). The co-authors are grateful and thank this Foundation and its Donors.

References