Characterization and DNA barcoding of Zambian plant species used as inoculum in the traditional fermentation of Munkoyo; a cereal-based beverage

Abstract

Background

Munkoyo, a non-alcoholic fermented beverage, is traditionally prepared in Zambia and neighbouring countries using cooked grains and the uncooked roots of wild plant species, collectively called ‘Munkoyo’ plants. The drink, valued for its refreshing taste and nutritional contribution, is made using roots of several wild plant species resulting in variations in the taste and quality of the beverage. However, comprehensive information on the specific plant species used in different regions of Zambia, as well as their occurrence in terms of habitat and soil type, is missing. This gap limits our understanding of the factors contributing to Munkoyo’s heterogeneity. The present study sought to identify the Zambian plant species used as an inoculum in Munkoyo fermentation and to characterize the soil in which they occur.

Results

Plant and soil samples were collected from four districts in Zambia known for Munkoyo production. Using morphological taxonomy, three Fabaceae species were identified as commonly used Munkoyo plants: Rhynchosia insignis (O.Hoffm.) R.E.Fr., Rhynchosia heterophylla Hauman, and Eminia holubii (Hemsl.) Taub. Root colour differed among these species, with the Rhynchosia species having yellowish roots and E. holubii having whitish roots. To validate their identification, we evaluated three DNA barcoding markers (matK, rbcL, and ITS2) for species discrimination. All markers showed 100% PCR amplification and sequencing success rates, with ITS2 displaying the highest genetic variability and species-level resolution. Phylogenetic analyses further confirmed ITS2 as the most effective marker. Validation using samples from a fifth district reaffirmed ITS2’s suitability for species-level discrimination. Soil analysis revealed significant associations between soil texture and plant occurrence: R. insignis and E. holubii were prevalent in sandy soils, while R. heterophylla was more prevalent in soils with lower sand content.

Conclusions

This study identified three common Munkoyo plant species and demonstrated ITS2 as a robust DNA barcode for their identification. It also established the influence of soil texture on the distribution of these plants, contributing to the understanding of Munkoyo production’s biological and environmental determinants.

Background

In many African countries where cereal crops are the staple food, traditional cereal-based fermented foods and beverages are a crucial part of diets for people, contributing to food and nutritional security [1]. During fermentation, the action of microorganisms and their enzymes convert the cereal ingredients into products with desirable characteristics such as enhanced organoleptic properties, improved nutritional value and longer shelf life [2]. For some of the fermented products, yeasts and lactic acid bacteria (LAB) present in the environment, processing equipment, food substrate, or added starter cultures are the dominant microorganisms involved in these fermentation processes [3]. However, our understanding of the raw materials and microbial communities driving traditional fermentation processes in most African products remains limited, impeding broader initiatives for large-scale production and commercialization [1, 3].

Munkoyo is a traditional non-alcoholic fermented beverage that is made from cereals and is widely consumed in Zambia and neighbouring regions [4]. The beverage is distinguished by its unique fermentation process, which uses uncooked roots from specific wild plants to introduce essential microbes and amylolytic enzymes to ferment cooked grains [2, 5]. To produce the beverage, debarked Munkoyo roots and/or root extract are added to a warm porridge made from cereal grains (maize, sorghum or millet) and allowed to ferment at ambient temperature for 24 to 48 h after which roots are sieved out and the beverage is ready for consumption [6]. While freshly prepared Munkoyo is traditionally consumed as a non-alcoholic beverage, prolonged fermentation can result in an increase in alcohol content [5]. Munkoyo is mostly prepared by women and consumed by people of all ages as a refreshing and nutritious drink. Additionally, it is used as a weaning beverage for infants as it is non-alcoholic. The beverage is also prepared on special occasions such as wedding ceremonies, funerals, church gatherings and traditional ceremonies. Apart from its appealing sweet-sour taste, improved nutritional value and prolonged shelf life, the local people also claim that the drink has health benefits, such as prevention and cure of diarrhoea or constipation [1, 6, 7].

Munkoyo is originally the local name for roots from plants belonging to the Fabaceae family, used in the production of the beverage as a fermentation starter [4]. These naturally occurring wild plants are classified under several genera, including Rhynchosia Lour., Eminia Taub., and Vigna Savi. They are indigenous to the Zambesian region, which comprises countries such as Zambia, the Democratic Republic of Congo, Zimbabwe, Angola, Tanzania, Malawi, Namibia, Botswana, and Mozambique [4, 6]. Typically, these shrub-like plants grow as understory vegetation in the Miombo Woodlands that span the Zambesian region.

Several plant species are reportedly used to produce Munkoyo beverage which according to consumers results in variations in the taste of the beverage. However, the underlying reason for the difference in the beverage can be manyfold as the plant species characteristics can differ depending on the field location where they were collected from and when buying the Munkoyo at the market it is hard to tell which exact plant species are being sold. The plant species themselves vary in ease of identification in the field, root colour, commercial value and other characteristics [4, 7]. While experienced individuals can distinguish Munkoyo species from other closely related and similar-appearing toxic species, the actual species used for Munkoyo production in various localities within the country, as well as their occurrence in terms of habitat and soil type, is unclear. Elucidating these aspects could lead to the improvement of the beverage through standardization and culturing of the preferred plant species with their associated microbes.

Traditional taxonomic morphological methods rely on examining whole plant features, particularly reproductive parts such as flowers, to accurately identify unknown plant species [8]. However, this identification process can be challenging when certain plant features are absent or when dealing with plants that have very similar characteristics. DNA barcoding, a molecular method that utilizes variations in short DNA regions (barcodes) to differentiate species, has emerged as an increasingly important tool for rapid plant identification, even in the absence of most plant parts or taxonomic expertise [9].

DNA barcoding, as recommended by the Consortium for the Barcode of Life (CBOL) Plant Working Group [17], is a powerful tool for species discrimination, and the selection of an appropriate barcode marker requires careful consideration of specific criteria outlined by Kress and Erickson [16]. Unlike in animals, where the gene encoding a mitochondrial cytochrome c oxidase I (COI) is a standard marker for species identification, this gene is not effective for plants due to the low rate of nucleotide substitution in plant mitochondrial genomes [10]. As a result, alternative markers have been proposed for plants. Commonly employed markers in previous studies include, the plastid genes like the Maturase-Kinase gene (matK) and ribulose-bisphosphate carboxylase/oxygenase large subunit (rbcL), as well as the internal transcribed spacer nuclear regions (ITS1 and ITS2). Combining these markers is vital to ensure accurate species identification [9, 11, 12].

Although DNA barcoding using matK, rbcL, and ITS2 markers has been employed to identify orchid species at risk of overharvesting in Zambia [13], no study has documented the use of DNA barcoding to identify Munkoyo plants, which are reportedly becoming scarce due to increased harvesting and deforestation [7]. Implementing DNA barcoding for the identification of Munkoyo plants used in various localities could facilitate the quick differentiation of distinct species that appear identical to inexperienced individuals. DNA barcoding could also contribute to reference databases for these plant species, which will be particularly useful for food authentication, traceability, and forensic botany applications, such as in cases of accidental death due to Munkoyo poisoning.

While previous studies have demonstrated that the microbial communities and amylolytic enzymes driving the fermentation process originate from Munkoyo roots, there is limited understanding of how abiotic soil properties and plant-inherent factors influence these microbial communities [5, 6]. Characterizing the soil properties in the natural habitat of Munkoyo plants provides a foundation for exploring the interplay between soil factors, root-associated microbes, and plant characteristics. This knowledge is crucial for understanding the natural factors that contribute to Munkoyo production and may ultimately inform strategies to enhance and standardize the fermentation process. Furthermore, insights into the soil habitat of Munkoyo plants could be vital for their successful domestication and cultivation.

In this study, we surveyed four districts in Zambia where Munkoyo plant species have been previously recorded [14], identifying three commonly used species through morphological taxonomy and examining their soil habitats. To further validate the identification process, plant samples were also collected from a fifth district, where DNA barcoding techniques were applied. The study aimed to achieve three specific objectives: [1] to identify the plant species used in Munkoyo production across selected districts in Zambia using both morphological and molecular methods; [2] to assess the effectiveness of three DNA markers, matK, rbcL, and ITS2, for molecular identification; and [3] to analyze the physicochemical properties of soils associated with the natural habitats of Munkoyo plants. These objectives were guided by the central research question: What are the specific plant species used in Munkoyo production, can they be accurately identified using molecular barcoding, and what soil characteristics define the habitats of these plants?

Results

Morphological characterization

All 15 plant specimens with accompanying vouchers (Supplementary Table 1, Additional file 1), belonged to the family Fabaceae and were distributed across two genera, Rhynchosia Lour. and Eminia Taub. Eight specimens were identified as Rhynchosia insignis (O.Hoffm.) R.E.Fr., which can be distinguished by their several erect stems with yellowish glandular hairs, 1–3 foliolate leaves with ovate leaflets, and yellow flowers with purple veins. Four specimens were identified as Rhynchosia heterophylla Hauman, which are characterized by their erect shrub with several erect shoots, sticky stems, 1–3 foliolate leaves, and yellow flowers. Finally, three specimens were identified as Eminia holubii Hemsl. Taub., which are characterized by their trailing or climbing stems growing from a woody rootstock, 3-foliolate leaves, and ovate to narrowly elliptic leaflets (see Fig. 1).

Regarding root characteristics, we observed that the Rhynchosia species had yellowish roots, while the roots of E. holubii had a whitish appearance after debarking.

Fig. 1

Morphological features of the sampled Munkoyo plant species. A) Rhynchosia insignis, characterized by 3 foliolate leaves and yellow flowers with purple markings. B) Rhynchosia heterophylla, distinguished by several erect stems, 3 foliate leaves and yellow flowers. C) Eminia holubii, characterized by trailing stems and 3 foliolate leaves. D), E), and F) show the roots of R. insignis, R. heterophylla, and E. holubii, respectively. R. insignis and R. heterophylla roots were yellowish, while E. holubii roots were white

Molecular characterization

PCR amplification efficiency and sequence characteristics

The success rates of PCR amplification and DNA sequencing are essential for evaluating DNA markers as barcodes. In our study, we found that the PCR amplification and sequencing efficiency for all three candidate barcodes were 100% for all 15 plant specimens. The matK DNA sequences ranged from 819 to 871 bp (mean 863 bp), with a mean GC content of 28.03%. For rbcL, the DNA sequences ranged from 573 to 599 bp (mean 582 bp), with a mean GC content of 43.09%. The ITS2 DNA sequences ranged from 350 to 466 bp (mean 433 bp), with a mean GC content of 49.32%.

Interspecific and intraspecific distance analysis

To evaluate the potential of the candidate DNA barcodes as reliable markers for Munkoyo species identification, we calculated several genetic distances using the Kimura 2-parameter model (K2P), as summarized in Table 1. Our analysis revealed that ITS2 exhibited the highest minimum interspecific genetic distance (0.0185), maximum intraspecific genetic distance (0.0033), and overall mean genetic distance (0.0804). This was followed by matK, which displayed a minimum interspecific genetic distance of 0.0011, a maximum intraspecific genetic distance of 0.0012, and an overall mean genetic distance of 0.0280. For rbcL, the minimum interspecific distance was 0.0000, the maximum intraspecific genetic distance was 0.0012, and the overall mean genetic distance of 0.0148.

Table 1 Genetic distance analysis of candidate DNA barcodes

Phylogenetic analyses

To further assess the potential of the three DNA markers as barcodes, we constructed maximum likelihood trees. The analysis revealed that all markers were able to distinguish the taxa at the genus level. The resulting trees exhibited two major clades representing the genera Eminia and Rhynchosia, indicating the genetic divergence between these two groups. Figures 2, 3 and 4 depict trees generated from the matK, rbcL, and ITS2 markers, respectively.

Fig. 2

Maximum Likelihood phylogenetic tree based on the matK gene. The tree exhibits two major clades, representing the genera Eminia (highlighted in green) and Rhynchosia (blue and red). The blue clade comprises Rhynchosia species from the current study, while the red clade represents Rhynchosia species retrieved from GenBank. Bootstrap values (> 50%) are indicated at the nodes, and a scale bar represents a genetic distance of 0.02. Acacia saligna was used to root the tree

Fig. 3

Maximum Likelihood phylogenetic tree based on the rbcL gene. The tree exhibits two major clades, representing the genera Eminia (highlighted in green) and Rhynchosia (blue and red). The blue clade comprises Rhynchosia species from the current study, while the red clade represents Rhynchosia species retrieved from GenBank. Bootstrap values (> 50%) are indicated at the nodes, and a scale bar represents a genetic distance of 0.005. To root the tree, Acacia baileyana F. Muell. was used

Fig. 4

Maximum Likelihood phylogenetic tree based on the ITS2. The tree exhibits several clades, each supported by high bootstrap values. These clades represent different species, with Eminia holubii highlighted in green, Rhynchosia insignis in brown, Rhynchosia heterophylla in blue, and the Rhynchosia species retrieved from GenBank in red. Bootstrap values (> 50%) are displayed at the nodes, and a scale bar indicates a genetic distance of 0.04. To root the tree, Acacia acuminata Benth. was used

Identification of additional plant samples using the DNA barcoding

To validate our DNA barcoding approach, we performed a BLAST search on NCBI using 13 DNA sequences from specimens collected in the fifth surveyed district (Mumbwa) and from other districts where morphology-based identification faced challenges due to the absence of voucher specimens (see Supplementary Table 2, Additional file 2). The BLAST results (summarized in Supplementary Table 3, Additional File 3) showed that the ITS2 marker provided highly accurate species-level identifications, with 11 sequences showing 100% identity to their closest matches in the GenBank database and two sequences showing 99.34% identity.

In contrast, the matK and rbcL markers, while yielding high sequence matches (100% identity), displayed redundancy in species-level resolution. For example, some sequences were matched equally to Rhynchosia insignis and Rhynchosia heterophylla, reflecting the inability of these markers to distinguish between closely related Rhynchosia species. We performed additional phylogenetic analyses by incorporating 13 samples that lacked accompanying vouchers. The resulting maximum likelihood trees (Supplementary Figs. 1, 2 and 3, Additional file 4) continued to show two distinct clades, with nine of the 13 samples clustering with the Rhynchosia species clade and the remaining four samples grouped with those identified as Eminia. Among these nine samples, ITS2 further separated the Rhynchosia clade into two sub-clades, with three samples clustering with R. heterophylla and six samples with R. insignis, supported by high bootstrap values.

Soil properties influence Munkoyo plant occurrence

The results of ANOVA and Tuckey’s tests revealed that the sampling sites where Munkoyo plants were growing differed significantly in their soil properties, including soil particle size (clay, sand, and silt), pH, organic matter (OM), electrical conductivity (EC), available phosphorus (available P), water retention at field capacity (FC) and permanent wilting point (PWP) (Supplementary Table 4, Additional file 5). To gain insights into the specific soil properties strongly associated with the occurrence of different Munkoyo plant species, we conducted canonical correspondence analysis (CCA). Before conducting CCA, we performed a Pearson correlation analysis (Supplementary Table 5, Additional file 6) to assess the relationships among soil variables. To ensure the reliability of our CCA results, we retained at least one variable from each pair of highly correlated variables (r > 0.8) in the CCA analysis (see Fig. 5). Specifically, we selected % OM, pH, % sand, and % clay while excluding EC, FC, PWP, and plant available water content (PAW) due to their high correlations with the selected parameters. Subsequent CCA ordination, using the selected variables and the dataset containing plant species presence-absence information, demonstrated the significance of the first four axes, as indicated by their eigenvalues (P < 0.05).

The CCA results showed that the variation in the data could be well explained by the first two axes in the CCA together explaining 76.6% of the variation. The first axis explained 56.3% of the variation and the second axis accounted for 20.3%. The first CCA axis was mainly determined by the % sand and clay in the soil and the Munkoyo plant species occurrence was separated along this axis, indicating a significant association between soil texture and the distribution of Munkoyo plant species. Specifically, R. insignis and E. holubii were more prevalent in sandy soils compared to R. heterophylla which was more associated with higher clay %. The second CCA axis was mostly determined by available P, pH and total nitrogen content (TN) in the soil, with E. holubii being associated with higher TN and lower pH and lower available P levels, whereas the Rhynchosia species showed the opposite pattern occurring in soils with higher pH, higher available P and lower TN.

Fig. 5

Canonical Correspondence Analysis (CCA) ordination plot showing the association between soil properties and the distribution of Munkoyo plant species. The arrows represent the correlation between soil variables and the position of each plant species on the ordination plot

Discussion

The primary objective of this study was to provide a comprehensive understanding of the identification and molecular characterization of Zambian plant species, whose roots initiate the fermentation process in the production of the traditional Munkoyo beverage. To achieve this, we used traditional morphological taxonomy and advanced molecular techniques, using three gene markers (matK, rbcL, and ITS2), to explore their potential as DNA barcodes for the precise identification of Munkoyo plant species through DNA Barcoding. Additionally, we analysed the main soil physicochemical properties at the sites where the Munkoyo plant species were growing to gain insights into the soil conditions that support the growth of these plants.

Morphological characterization

Using morphological taxonomic methods, our study identified three commonly used plant species: Rhynchosia insignis, Rhynchosia heterophylla, and Eminia holubii, all belonging to the Fabaceae family. These findings are in line with earlier reports indicating that plants used to produce Munkoyo in Zambia and surrounding areas include members of the genera Rhynchosia, Eminia, and Vigna [4, 7]. However, we did not encounter any use of Vigna species in the selected areas, and these species have been particularly reported for use in the Democratic Republic of Congo [4]. Different tribes and cultures might use certain species specific to their geographic regions, a hypothesis that remains to be determined. Future surveys, including other sampling areas, may identify additional plant species.

Furthermore, morphological characterization revealed clear distinctions among the identified plant species. A particularly prominent difference was observed in root colouration: the roots of both Rhynchosia species exhibited a yellowish hue, while Eminia holubii displayed whitish roots. These findings corroborate previous reports of morphological variation, especially in root appearances, among Munkoyo plant species [7]. Additionally, the observed variations support the relevance and reliability of traditional taxonomic methods in identifying plant species used in Munkoyo production.

Despite these advances, our findings highlight the need for further surveys to explore additional regions and potentially identify other plant species used in Munkoyo production. Such studies could provide deeper insights into the geographic and cultural factors influencing the selection of specific plant species, ultimately contributing to a more comprehensive understanding of the diversity of Munkoyo plant resources.

Molecular characterization

In this study, molecular techniques were used to characterize the selected Munkoyo plant species, with the primary objective of evaluating the suitability of three candidate DNA barcodes (matK, rbcL, and ITS2) for accurate species identification. High PCR amplification and sequencing success rates were observed, with all three markers achieving 100% efficiency across the 15 plant specimens analyzed. These results show their potential applicability for Munkoyo plant identification and align with previous studies demonstrating that DNA sequences can reliably be obtained from diverse tropical plant species using these primers [14].

An essential criterion for a robust DNA barcode is its ability to distinguish species based on significant genetic variability and divergence. Among the markers tested, ITS2 demonstrated the highest genetic variability and species-level discrimination, proving to be the most effective for differentiating the three identified Munkoyo plant species. These findings are consistent with earlier studies that have highlighted ITS2’s utility in plant species identification across various families, including Fabaceae [11, 15,16,17]. However, it is important to acknowledge that the effectiveness of ITS2 as a barcode can vary among taxonomic groups, as shown in families like Brassicaceae and Rosaceae, where its resolution is limited [10, 17].

The phylogenetic analyses further validated the suitability of ITS2 as a barcode for Munkoyo plant species identification. While matK and rbcL provided limited resolution, grouping the samples into two major clades, ITS2 achieved finer resolution by distinguishing the three species into distinct clades that corresponded with morphological identifications. This high discriminatory power highlights ITS2’s potential for precise identification of closely related species within the same genus, a crucial aspect for understanding the diversity and application of Munkoyo plants.

Additionally, BLAST analyses reinforced the effectiveness of ITS2 as a barcode, accurately identifying species that could not be distinguished morphologically. Interestingly, the vernacular names traditionally used for these plants aligned closely with the species identified through ITS2 phylogenetic trees (Supplementary Figs. 1, 2, and 3; Additional File 4). For instance, the names Munkoyo and Chitondo corresponded to Rhynchosia insignis, Kailunge and Kelunge to Rhynchosia heterophylla, and Chipuma, Mupuma, and Mulaba were consistently associated with Eminia holubii.

The accurate identification of Munkoyo plant species is vital for several reasons. First, it lays the foundation for further research into how different species influence the composition and quality of the beverage, thereby supporting the development of standardized production practices. Second, precise identification methods, such as DNA barcoding, can be applied across various regions, offering a broader understanding of local plant utilization. Additionally, molecular identification aids in building reference databases essential for food authentication, traceability, and forensic botany. This is particularly critical in addressing cases of Munkoyo poisoning, where toxic look-alike species may be inadvertently used instead of the non-toxic plants traditionally employed.

Occurrence and soil associations of Munkoyo plant species

This study investigated the occurrence of Munkoyo plant species and their associations with soil properties across various habitats. Among the three identified species, R. insignis was the most commonly occurring, while R. heterophylla was the least common. Notably, co-occurrence of all three species was rare, observed only at one site in Kasempa district.

The analysis revealed a strong association between soil texture and the distribution of Munkoyo species. Specifically, R. insignis and E. holubii were predominantly found in sandy soils, whereas R. heterophylla was more prevalent in clay-rich environments. This aligns with prior studies reporting the occurrence of E. holubii on Kalahari desert sands, supporting the species’ preference for such environments [18]. Beyond texture, the soil’s chemical properties also influenced species distribution. CCA analysis revealed distinct preferences among the species. E. holubii was associated with soils high in total nitrogen (TN), low in pH, and low in available phosphorus (P), while R. insignis and R. heterophylla favoured soils with higher pH, increased available P, and lower TN levels. However, these findings represent associations rather than definitive preferences. To confirm these patterns, further experimental studies are needed to test plant growth across a range of controlled soil conditions. Such research would clarify whether the observed distributions are due to inherent soil preferences or influenced by other ecological factors, such as dispersal limitations or competition.

The rising demand for Munkoyo, coupled with the overharvesting of its wild plant sources, underscores the need for domestication and cultivation strategies. Understanding the soil requirements of these plants is essential for successful cultivation. While soil texture and chemistry are critical, microbial communities within these soils likely play an equally important role. These microbes not only influence plant growth but also contribute to the fermentation process, impacting the quality of the Munkoyo beverage [5, 6]. Efforts to cultivate Munkoyo plants should thus adopt a multidisciplinary approach. Beyond replicating soil conditions, cultivation practices should consider the microbial ecology of these environments. Maintaining the native microbial communities associated with these plants can help preserve the distinctive qualities of traditional Munkoyo while ensuring sustainable production.

Conclusions

In this study, we conducted a comprehensive investigation into the plant species used in the production of Munkoyo, a popular non-alcoholic beverage in Zambia. We combined traditional morphological taxonomy with advanced molecular techniques, successfully identifying three commonly used Munkoyo plant species while highlighting the potential of ITS2 as a robust DNA barcode for their precise identification. It also established the influence of soil texture on the distribution of these plants, contributing to the understanding of Munkoyo production’s biological and environmental determinants. Additionally, our research unveiled a compelling connection between soil properties, particularly texture, and the distribution of these plant species.

Materials and methods

Study area

The study was conducted across four districts in Zambia: Nkeyema in the Western Province, Rufunsa in Lusaka Province, Kasempa in the North-Western Province, and Mkushi in the Central Province, where detailed investigations were carried out. Additionally, Mumbwa in the Central Province was included for plant sampling to complement the data from the primary study sites. These districts were chosen based on the presence of Munkoyo plants (Fig. 6).

Fig. 6

Geographical map of Zambia showing the sampling locations. Map created by Mubonda Kalumbilo and Christopher Lusekelo-2023

Plant sampling and morphological identification

Sampling was carried out from December 2020 to November 2021. In each district, knowledgeable local community members, familiar with the traditional use and identification of Munkoyo plants, were actively involved in identifying these plants directly in the field. For morphological characterization, at least two voucher specimens for each plant species that exhibited distinct morphological characteristics were collected in a plant press and transported to the University of Zambia herbarium (code: UZL) for taxonomic identification and storage. The plants were identified using prior knowledge and comparisons with herbarium reference specimens. To confirm species identifications, the collected specimens and accompanying photographs were also compared with online repositories of herbarium specimens, namely the Flora of Zambia [19] and JSTOR Global Plants [20]. Morphological features were carefully examined and compared to the available descriptions and images in these repositories. Plant identifications were verified by David Chuba.

Molecular characterization

For molecular characterization, we selected 15 plant specimens that were accompanied by vouchers. Each plant species included at least three individual replicates collected in at least two different districts, with their identity being determined by their morphology (Supplementary Table 1, Additional file 1). During field sampling fresh Munkoyo leaves, weighing approximately 20 g, were collected in sterile zip-lock bags and immediately dried using silica gel until DNA extraction.

DNA extraction, PCR amplification and sequencing

To extract total genomic DNA from each sample, 100 mg of silica-dried leaves were used. The tissues were ground in liquid nitrogen before DNA was isolated using a DNeasy Plant Pro kit (Qiagen, Germany) according to the manufacturer’s instructions, with a final elution volume of 50 µl. The isolated DNA was quantified by spectrophotometry using a Nanodrop (ND-2000) Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Extracted DNA was stored at -20 °C until further analysis.

Polymerase chain reaction (PCR) was carried out using universal primers for the two plastid markers (rbcL and matK) and the nuclear region ITS2 (Table 2). Genomic DNA ranging from 20 to 150 ng was used in the PCR. The reactions were performed in a total reaction volume of 25 µl containing 17.375 µl of nuclease-free water, 5.0 µl PCR Buffer 5 X, 0.5 µl 10 µM dNTP mix, 0.5 µl each of 10 µM primer, 0.125 µl Taq DNA polymerase and 1.0 µl of genomic DNA. The PCR program consisted of an initial denaturation at 95 °C for 2 min, followed by 35 cycles of denaturation at 95 °C for 1 min, annealing at 50 °C for rbcL gene, 55 °C for matK gene and 56 °C for ITS2, for 1 min, elongation at 72 °C for 1 min and a final extension at 72 °C for 5 min. The amplified PCR products were visualized on 1% agarose gels stained with ethidium bromide visualized under UV light using the BIORAD Molecular Imager^® Gel Doc™ XR Imaging System (BIORAD, USA). The PCR products were sequenced bidirectionally using an ABI 3730XL automated sequencer (Eurofins Genomics, Germany).

Table 2 Primers used in the study

Sequence alignment and data analysis

The reads obtained from forward and reverse sequences of all PCR products were analyzed using Geneious R11 [22]. To ensure sequence quality, the peaks corresponding to each nucleotide were scrutinized, and segments of poor-quality DNA sequence ends were trimmed. Subsequently, the forward and reverse sequences were aligned to generate consensus sequences, which were exported to molecular evolutionary genetic analysis (MEGA11) software [23] for further analysis.

Genetic distance and phylogenetic analysis

To evaluate the potential of the candidate DNA barcodes as reliable markers for identifying Munkoyo species, we employed the K2P model to estimate the nucleotide divergence between sequences. Each barcode’s corresponding consensus sequences were aligned individually using the MUSCLE algorithm within MEGA11. To evaluate the level of genetic differentiation within and between Munkoyo species, we then calculated the overall mean genetic distances, as well as the intraspecific and interspecific genetic distances.

For phylogenetic analyses, we conducted a Basic Local Alignment Search Tool (BLAST) search to identify the closest matching sequences in the National Centre for Biotechnology Information (NCBI) GenBank [24] for each of the three candidate DNA barcodes. Although we did not find exact matches for our sequences, we retrieved several sequences with high similarity (> 90%) that belonged as morphologically determined, to at least the same genera as our samples. From these, we selected one representative sequence for each different species within the same genera. As outgroups to root the phylogenetic tree, we used Acacia species from a different subfamily (Mimosoideae) of Fabaceae, based on our morphological identifications. Maximum likelihood phylogenetic trees were generated for each gene using the K2P model with a “Bootstrap phylogeny” value set at 1000 in the MEGA11 software. The maximum likelihood trees were edited and visualized using FigTree software (version 1.4.4) [25]. All sequences were deposited in NCBI GenBank.

To further validate the effectiveness of the candidate DNA barcodes in identifying Munkoyo plant species, we conducted additional phylogenetic analyses by incorporating DNA sequences from Mumbwa district and plant specimens that were initially excluded from the molecular characterization process due to the lack of accompanying voucher specimens. In parallel, we performed a BLAST search on NCBI for these sequences to compare them against the GenBank database, assessing their similarity to known species.

Soil sampling and analysis

Soil sampling was undertaken at the same time as plant sampling to assess the soil habitat of the Munkoyo plants. The size of the sampling locations varied across the districts, ranging from 20 to 50 m in radius. At each surveyed location where the plant specimens were collected, we obtained soil samples from a distance of 1–2 m away from at least three Munkoyo plants. To ensure representative sampling, the selected plants in each location were growing at least 10 m apart, resulting in a total of three soil samples per site. Supplementary Table 6, Additional File 7 contains specific sampling location details, including site descriptions, while Additional File 8 presents a plant and soil sampling map, depicting how the sampling was conducted.

Before soil collection, we removed the upper layer of vegetation using a shovel and sampled approximately 500 g of the 0–30 cm soil layer using a garden trowel. Each soil sample was then carefully placed in a zip-lock bag to ensure proper preservation during transportation. The collected soil samples were subsequently transferred to the University of Zambia Agric-Soil Science Laboratory for further analysis.

The soil samples were air-dried and sieved using a 2 mm sieve in preparation for various physical and chemical analyses. We analyzed several physicochemical parameters, including soil texture, soil pH, organic matter (OM) content, water holding capacity (WHC), available phosphorus (available P) content, total nitrogen content (TN), and electrical conductivity (EC).

Water holding capacity was determined by measuring the soil water content at field capacity (FC) and permanent wilting point (PWP) using a pressure membrane extractor system [26]. The plant available water content (PAW) was calculated as the difference between the WHC and PWP. The particle size distribution of the soil samples was determined using the hydrometer method [27]. The TN content of the soil was determined using the micro-Kjeldahl method [28] and OM content using the Walkley and Black method [29]. Available P was determined using the Bray 1 method [30]. The soil pH and EC were measured using a pH meter and an EC meter, respectively.

Statistical analysis

To investigate the relationship between soil properties and the occurrence of Munkoyo plants, we compared the soil properties across the sampling locations and noted the presence or absence of the different Munkoyo plant species. The analysis of the different soil parameters included a one-way analysis of variance (ANOVA) with sampling location as the predicting variable and each of the main soil characteristics as the response variable, which was performed in Minitab version 17.0 software. Furthermore, we conducted a Canonical Correspondence Analysis (CCA) using Canoco (version 5.1) software [31] to gain insights into the specific soil properties that were most strongly associated with the occurrence of the different Munkoyo plant species. CCA is a multivariate statistical technique that allows for the examination of species-environment relationships [32]. To determine the significance of these plant species-environment correlations, we performed a Monte Carlo permutation test with 499 permutations [33].