The Fusarium oxysporum species complex (FOSC) is widely recognized as one of the most economically important groups of plant-pathogenic fungi and is a taxonomically problematic group within the genus Fusarium (Dean et al., 2012; Laurence et al., 2014; Lombard et al., 2019). Members of this complex are globally distributed and exhibit diverse ecological roles, including saprotrophic, endophytic, phytopathogenic, and opportunistic pathogenic lifestyles in animals and humans (Constantin et al., 2021; O’Donnell et al., 2010; Todorovic et al., 2023). Despite its importance in agriculture, ecology, and biosecurity, species delimitation within FOSC remains unresolved and contentious.
Traditionally, species classification within F. oxysporum has relied largely on morphological characters and host specificity, with plant-pathogenic strains restricted to specific plant hosts or narrow host ranges being designated as formae speciales (Edel-Hermann and Lecomte, 2019). By early 2025, approximately 130 F. oxysporum formae speciales had been reported worldwide (Han et al., 2025). However, accumulating evidence indicates that the FOSC comprises multiple morphologically cryptic species (Achari et al., 2020; Lombard et al., 2019), and many isolates assigned to the same forma specialis are polyphyletic, while closely related or even conspecific isolates may belong to different formae speciales (Baayen et al., 2000; Han et al., 2025). This conflict reflects the fact that host specificity can evolve repeatedly and independently within the complex, and therefore does not reliably indicate shared evolutionary history. Consequently, reliance on morphological traits and host-fungus associations alone is insufficient for robust species delimitation within the FOSC.
Commonly used loci for species delimitation within FOSC include the translation elongation factor 1-alpha (tef1), RNA polymerase II subunits (rpb1, rpb2), calmodulin (CaM), and beta-tubulin (tub2) (Achari et al., 2020; Wang et al., 2022; Zhang et al., 2024). These loci often provide insufficient resolution when analyzed individually and are problematic in combined analyses, as individual gene phylogenies frequently exhibit incongruent topologies, weak support for internal nodes, and conflicting placements of closely related isolates (Achari et al., 2020; Brankovics et al., 2020). Such discordance is commonly attributed to incomplete lineage sorting, recombination, and potential gene flow among lineages within the FOSC (Fourie et al., 2009; Laurence et al., 2012; McTaggart et al., 2021). Moreover, the recent multilocus phylogenetic analyses (Han et al., 2025; Wang et al., 2022; Zhang et al., 2024) indicate that several lineages/species within FOSC appear frequently near the threshold of speciation, raising uncertainty as to whether they represent distinct species or structured populations within a single species. These difficulties highlight the inadequacy of relying on individual gene trees or arbitrary genetic distance thresholds for species recognition in this complex.
The Genealogical Concordance Phylogenetic Species Recognition (GCPSR) concept provides an evolutionary framework for species delimitation based on multilocus genealogies (Taylor et al., 2000). Under this framework, species boundaries are inferred from the transition between genealogical concordance among loci and genealogical conflict within species. Subsequent refinements have operationalized GCPSR by establishing criteria for recognizing independent evolutionary lineages and assigning them to species (Brankovics et al., 2017; Dettman et al., 2003a, 2003b; Glassnerova et al., 2022). GCPSR has been widely applied in fungal systematics, including within the FOSC, where it has facilitated the recognition of cryptic diversity and improved taxonomic resolution (Brankovics et al., 2017; Laurence et al., 2014). However, applying GCPSR for fungal species delimitation can be challenging due to gene tree discordance, which may obscure true evolutionary relationships among closely related taxa (Achari et al., 2020; Liu et al., 2016).
In recent years, multispecies coalescent (MSC)-based approaches have provided a statistically rigorous framework for species delimitation in groups affected by incomplete lineage sorting and recent divergence (Degnan and Rosenberg, 2009; Maharachchikumbura et al., 2021). Among MSC-based approaches, STACEY (Jones, 2017) has been increasingly adopted in fungal systematics and is widely used for species delimitation in several fungal genera such as Aspergillus, Penicillium and other Eurotiales (Bian et al., 2022; Glassnerova et al., 2022; Sklenar et al., 2022; Visagie et al., 2025), as well as the F. sambucinum species complex (Sandoval-Denis et al., 2025). STACEY often yields more biologically meaningful and stable species boundaries than distance-based and tree-based single-locus methods such as Assemble Species by Automatic Partitioning (Puillandre et al., 2021), Automatic Barcode Gap Discovery (Puillandre et al., 2012), the generalized mixed Yule-coalescent (Fujisawa and Barraclough, 2013), Bayesian implementation of the general mixed Yule-coalescent (Reid and Carstens, 2012), Poisson tree processes and Bayesian Poisson tree processes (Zhang et al., 2013), which can vary substantially depending on the locus analyzed or parameter settings (Glassnerova et al., 2022; Sandoval-Denis et al., 2025). Although the MSC-based methods provide a powerful framework for species delimitation, their performance is strongly influenced by locus selection, sampling design, population structure, and underlying biological assumptions (Chambers and Hillis, 2020; Fujita et al., 2012).
Many effective approaches have been developed and widely applied in fungal taxonomy. However, there is no single gold standard method that can be consistently applied across all fungal groups, as species boundaries vary considerably among lineages and datasets (Lucking et al., 2020; Stengel et al., 2022). Therefore, species delimitation should rely on integrative approaches, including morphological, ecological, multilocus phylogenetic, genome-scale, and coalescent-based evidence, to achieve more reliable taxonomic conclusions in fungi (Cao et al., 2021; Maharachchikumbura et al., 2021; Stengel et al., 2022; Zhang et al., 2026).
Fusarium oxysporum is one of the most important plant pathogens in Korea, causing a wide range of diseases such as wilt, damping-off, and root and stem rot across diverse host plants (KSPP, 2024). A large number of F. oxysporum sensu lato (s. l.) isolates have been preserved in the Korean Agricultural Culture Collection (KACC), representing diverse geographic regions and substrates since the 1990s. Most of these isolates were originally identified as F. oxysporum s. l. and its formae speciales, primarily based on morphological characteristics and host/substrate association. However, such criteria are often insufficient for accurate species delimitation within this complex. Re-identification of these isolates is therefore essential for improving our understanding of species diversity within the FOSC, as well as for enhancing disease management strategies and supporting plant quarantine efforts in Korea. Moreover, Korean isolates represent a valuable resource for providing critical data to refine species boundaries within this group.
Given these unresolved taxonomic issues, this study aims to (i) reassess species boundaries within FOSC using multilocus phylogenetic analyses, the GCPSR concept, and multispecies coalescent-based species delimitation approaches and (ii) re-identify Fusarium isolates from Korea preserved in KACC and re-evaluate the species diversity of FOSC in Korea.
Materials and Methods
Fungal isolates
A total of 92 Fusarium isolates preserved in KACC were examined in this study. These isolates were collected from diverse hosts and substrates in Korea since the 1990s and were preliminarily identified as members of the FOSC based on tef1 sequence similarity analyses. Detailed information for all isolates included in this study is provided in Table 1.
DNA extraction, PCR amplification and sequencing
Isolates were cultured on potato dextrose agar (PDA; Difco Laboratories, Detroit, MI, USA) at 25°C in the dark for 5 days. Fresh mycelial biomass was harvested from the colony margins and used for genomic DNA extraction. Total genomic DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions.
Five loci commonly used in Fusarium phylogenetic studies were amplified and sequenced, including tef1, rpb1, rpb2, CaM, and tub2. PCR amplification and sequencing were performed using the primer pairs EF-1/EF-2 for tef1 (O’Donnell et al., 1998), CL1/CL2A for CaM (O’Donnell et al., 2000), T1/T2 for tub2 (O’Donnell and Cigelnik, 1997), and 5F2/7cR for rpb2 (Liu et al., 1999; Sung et al., 2007). For rpb1, the primer pairs Fa/R8 and F7/G2Rm were used for PCR amplification. Amplicons generated with Fa/R8 were sequenced using the internal primers F5 and R8, whereas amplicons generated with F7/G2Rm were sequenced using the same primer pair (O’Donnell et al., 2010; Thao et al., 2025). The PCR primers were listed in Supplementary Table 1.
PCR reactions were performed in a total volume of 25 μL, consisting of 12.5 μL of 2× PCR Master Mix (MG Med, Korea), 8.5 μL of nuclease-free water, 1 μL of each primer (4.5 pmol), and 2 μL of genomic DNA (100 ng/μL). Amplifications were conducted using an AllInOneCycler Thermal Cycler (Bioneer, Daejeon, Korea) under cycling conditions described by Thao et al. (2025) (Supplementary Table 1).
Amplified products were purified using the QIAquick PCR Purification Kit (Qiagen) and verified by agarose gel electrophoresis. Purified PCR products were subsequently submitted to Macrogen (Seoul, Korea) for bidirectional Sanger sequencing.
Phylogenetic analyses
Forward and reverse sequence reads were edited, assembled, and trimmed using MEGA v. 11 (Tamura et al., 2021). Newly generated sequences were deposited in GenBank. Representative sequences of previously described species within the FOSC were retrieved from the FUSARIOID-ID database and GenBank for comparative analyses. Fusarium fujikuroi CBS 221.76 was selected as the outgroup taxon.
Sequences of each locus were aligned separately using MAFFT v. 7 (Katoh and Standley, 2013) under the automatic alignment strategy and manually adjusted where necessary. Individual locus alignments were subsequently concatenated in MEGA for multilocus phylogenetic analyses.
Phylogenetic reconstructions were performed using both single-locus and concatenated datasets. Maximum likelihood (ML) analyses were conducted using IQ-TREE v. 3.0.1 (Nguyen et al., 2015). The optimal nucleotide substitution model for each partition was determined using ModelFinder (Kalyaanamoorthy et al., 2017) according to the Bayesian Information Criterion. Branch support was estimated using 1,000 ultrafast bootstrap replicates. The resulting ML trees were visualized in MEGA.
Bayesian inference (BI) analyses were performed in MrBayes v. 3.2.7 (Ronquist et al., 2012). Substitution models for each partition were selected using ModelFinder in IQ-TREE and subsequently adapted to the closest compatible model parameters implemented in MrBayes. Two independent runs, each consisting of four Markov chains, were conducted for 10,000,000 generations with trees sampled every 1,000 generations. Convergence of the analyses and effective sample size (ESS) values were assessed using Tracer v. 1.7.2 (Rambaut et al., 2018), and ESS values above 200 were considered indicative of sufficient sampling. The first 25% of sampled trees were discarded as burn-in. Consensus trees were visualized using FigTree v. 1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/). Alignments and phylogenetic trees were deposited in Figshare (https://doi.org/10.6084/m9.figshare.32671977).
Hypothesized evolutionary lineage delimitation analyses
To investigate putative evolutionary lineages within the FOSC and evaluate alternative lineage boundary hypotheses, multiple complementary analytical approaches were applied, including exploratory multilocus coalescent clustering using STACEY, GCPSR, and multispecies coalescent validation using Bayesian Phylogenetics and Phylogeography (BPP).
To reduce computational redundancy and minimize the influence of duplicate multilocus genotypes, redundant haplotypes were removed prior to GCPSR and coalescent-based analyses. The concatenated multilocus alignment was imported into R v. 4.5.2 (R Core Team, 2025), and identical haplotypes were identified using the haplotype function implemented in the package pegas (Paradis, 2010). Only unique haplotypes were retained, and the resulting reduced dataset was subsequently partitioned again into the corresponding individual loci for downstream analyses.
Exploratory lineage clustering using STACEY
Exploratory lineage clustering analyses were conducted using STACEY v. 1.3.1 (Jones, 2017) implemented in BEAST v. 2.7.7 (Bouckaert et al., 2019). The XML configuration file was generated using BEAUti. The optimal substitution model for each locus, determined by ModelFinder, was applied separately for each partition. A strict molecular clock model was applied to all loci. The species tree prior followed the STACEY birth-death collapse model. The collapseheight parameter was initially set to 1.0 × 10−4. Additional priors included a lognormal prior for bdcGrowthRate (M = 5, S = 2), a beta prior for collapseWeight (α = 1, β = 1), a lognormal prior for popPriorScale (M = −7, S = 2), and a beta prior for relativeDeathRate (α = 1, β = 1000) (Visagie et al., 2025). Markov chain Monte Carlo (MCMC) analyses were run for 9 × 108 generations.
Convergence diagnostics and parameter sampling were evaluated using Tracer v. 1.7.2 (Rambaut et al., 2018) after discarding the first 20% of samples as burn-in, with all ESS values exceeding 200. Posterior clustering distributions were summarized using SpeciesDA (Jones, 2017). To assess the sensitivity of species delimitation to the collapseheight parameter, species delimitation outcomes and the posterior probabilities of the two most frequently recovered clustering scenarios were examined across a range of collapseheight values following Sklenar et al. (2022). Collapseheight values for subsequent analyses were selected from stable parameter intervals in which the inferred species delimitation scenario remained unchanged across consecutive collapseheight values. Posterior similarity matrices were then generated from the corresponding SpeciesDA outputs and visualized as heatmaps following the framework of Jones et al. (2015). Species delimitation scenarios inferred under different collapseheight values were subsequently mapped onto and compared on a phylogenetic tree inferred from the concatenated dataset. Files associated with the STACEY species delimitation analyses have been deposited in Figshare (https://doi.org/10.6084/m9.figshare.32671977).
Genealogical concordance phylogenetic species recognition (GCPSR)
Genealogical concordance analyses were performed following the GCPSR framework proposed by Dettman et al. (2003a, 2003b) and subsequently refined by Glassnerova et al. (2022), with minor modifications. The reduced datasets were used to construct individual-locus trees (Supplementary Figs. S1-S6) and a combined tree for GCPSR assessment.
A clade was considered to represent an independent evolutionary lineage when it satisfied either the criterion of genealogical concordance or genealogical non-discordance. Under the genealogical concordance criterion, a clade was required to be recovered in at least two-thirds of the single-locus phylogenies. Under the genealogical non-discordance criterion, a clade had to receive strong statistical support in at least one single-locus phylogeny (≥70% ML bootstrap support and ≥0.95 Bayesian posterior probability) while not being contradicted with comparable support in any other locus.
To determine which evolutionary lineages should be recognized as phylogenetic species, two additional criteria were assessed using the multilocus phylogeny. First, candidate species were required to exhibit clear genetic differentiation from closely related taxa, thereby avoiding the recognition of small terminal clades as independent species. Second, all sampled isolates were required to be assignable to a phylogenetic species so that no isolates remained unclassified.
Multispecies coalescent validation using BPP
Multispecies coalescent analyses were performed using BPP v. 4.8.7 (Flouri et al., 2018) to evaluate alternative lineage hypotheses inferred from phylogenetic analyses, STACEY clustering, and GCPSR. Analyses focused on the F. inflexum lineage and several closely related taxa, including F. tardichlamydosporum, F. vanleeuwenii, F. libertatis, and F. irdabamae.
Two alternative lineage hypotheses were tested. The first represented a fine-scale delimitation scenario recognizing 17 putative lineages, whereas the second represented a conservative scenario recognizing five broader putative lineages (Supplementary Fig. 7).
Multispecies coalescent validation analyses were conducted under both the A10 and A11 models in BPP. Under the A10 model, the proposed lineage hypotheses were evaluated using the species validation framework with a fixed guide tree inferred from the multilocus maximum likelihood phylogeny. Under the A11 model, lineage hypotheses and the species-tree were jointly inferred using the same topology as the starting tree. In both analyses, reversible-jump Markov chain Monte Carlo sampling was employed to explore alternative species delimitation models. Fine-tuning parameters were automatically adjusted using the default settings recommended in BPP.
To assess the sensitivity of species delimitation to prior assumptions, three combinations of gamma priors for ancestral population size (θ) and root age (τ0) were tested following Hubka et al. (2018): (i) large ancestral population size with deep divergence, θ ~ G(1, 10) and τ0 ~ G(1, 10); (ii) small ancestral population size with shallow divergence, θ ~ G(2, 2000) and τ0 ~ G(2, 2000); and (iii) large ancestral population size with shallow divergence, θ ~ G(1, 10) and τ0 ~ G(2, 2000).
Each prior combination was replicated three times using independent runs. For all analyses, two independent MCMC chains were run for 500,000 generations with a burn-in of 32,000 generations, and samples were collected every two generations. Convergence among runs was assessed by comparing posterior probabilities across independent analyses. Posterior probabilities were used to evaluate support for alternative lineage hypotheses under the multispecies coalescent framework.
Literature survey of FOSC diversity in Korea
Published records of FOSC species and their host/substrate associations in Korea were compiled from peer-reviewed articles, the List of Plant Diseases in Korea (KSPP, 2024), the National List of Species of Korea (2024) (https://kbr.go.kr), and the USDA Fungal Databases (https://fungi.ars.usda.gov/).
Results
Phylogenetic analyses
The dataset comprised 263 strains and 1,088 DNA sequences generated from five loci, including tef1 (263 sequences), rpb1 (182), rpb2 (255), CaM (203), and tub2 (185). Among these, 626 sequences representing 49 previously recognized taxa were retrieved from GenBank, whereas 457 sequences were newly generated from 92 Korean isolates and five sequences were obtained from the outgroup strain F. fujikuroi CBS 221.76.
The concatenated five-locus alignment comprised 3,984 characters, including gaps. Of these, 3,359 characters were conserved, 624 were variable, and 345 were parsimony-informative. The alignment partitions corresponded to tef1 (positions 1-623), rpb1 (624-2,048), rpb2 (2,049-2,911), CaM (2,912-3,495), and tub2 (3,496-3,984).
Phylogenetic analyses based on both ML and BI recovered broadly congruent topologies. The multilocus phylogeny (Fig. 1) resolved the 262 strains within the FOSC into 32 clades, representing the 32 phylogenetic species recognized in this study. Among these, 16 taxa were nested within clades representing previously recognized species: F. liaoningense and F. vaughaniae were embedded within the F. cugenangense clade with strong support (ML bootstrap support/Bayesian posterior probability = 99/0.99); F. hordeum within the F. agrestense clade (100/1); F. fabacearum within the F. inflexum clade (94/0.93); F. sangayamense within the F. kalimantanense clade (100/1); F. moutoniae, F. pharetrum, and F. veterinarium within the F. contaminatum clade (100/1); F. coriorum within the F. curvatum clade (100/1); F. houttuyniae, F. idabrowneae, F. landiae, and F. liriopes within the F. nirenbergiae clade (100/1); and F. cili, F. phialophorum, and F. rosae-roxburghii within the F. odoratissimum clade (99/1). However, F. elaeidis was placed within the F. grosmichelii clade with only moderate ML support and without significant BI support (72/<0.90).
Hypothesized evolutionary lineages within the FOSC. Lineages inferred from STACEY
A total of 263 sequences from the concatenated alignment were reduced to 198 unique sequences using the haplotype function. The reduced dataset was subsequently partitioned into separate single-locus alignments for STACEY analyses.
Across a broad range of collapseheight values, posterior support was distributed across multiple alternative clustering schemes, and even the most frequently recovered solutions generally received low posterior probabilities (Supplementary Fig. 8). These results suggest limited genealogical differentiation among many lineages within the FOSC and indicate that species boundaries remain difficult to resolve confidently under the multispecies coalescent framework.
Several collapseheight values corresponding to relatively stable clustering outcomes were selected for further comparison, including 0.00151, 0.00191, 0.00239, and 0.00285. These settings recovered clustering scenarios ranging from 11 to 21 inferred lineages (Fig. 2; Supplementary Figs. 9-12). Among them, the clustering schemes obtained at collapseheight values of 0.00151 and 0.00191 showed the highest overall concordance with the multilocus phylogeny and the subsequent GCPSR and BPP analyses, suggesting a tentative 18-lineage scenario (highlighted with boxes in Fig. 2). Notably, the largest inferred lineage, here referred to as the F. inflexum clade, encompassed 18 previously recognized species. In addition, the F. languescens lineage, comprising F. languescens and F. queenslandicum, and the F. nashariae lineage, comprising F. nashariae and F. paraodoratissimum were also inferred.
Genealogical concordance phylogenetic species recognition (GCPSR)
Because the tub2 locus exhibited substantial phylogenetic incongruence and limited discriminatory power across the FOSC (Supplementary Fig. 5), this marker was excluded from GCPSR analyses. Consequently, lineage assessment under GCPSR was based on the remaining loci (tef1, rpb1, rpb2, and CaM) (Supplementary Figs. 1-4).
The results of the GCPSR analyses are summarized in Fig. 2, in which 19 putative evolutionary lineages within the FOSC were recognized. One of the most notable results involved the F. inflexum clade, a putative evolutionary lineage comprising numerous ex-type and reference strains previously described as distinct species, including F. vaughaniae, F. cugenangense, F. liaoningense, F. carminascens, F. hordeum, F. agrestense, F. gossypinum, F. inflexum, F. fabacearum, F. grosmichelii, F. elaeidis, F. menglaense, F. hexaseptatum, F. amerosporum, F. scabridum, F. glycines, F. wimaladesilvae, and F. duoseptatum.
The F. inflexum lineage was supported in the tef1 genealogy (ML/BI = 90/0.96) and was not contradicted with comparable support in any other locus. Although several well-supported subclades were observed within this lineage, their composition varied among loci, and no subdivision was consistently recovered across genealogies. These conflicting topologies (Supplementary Fig. 6) suggest that the observed internal structure likely reflects population-level variation rather than independently evolving lineages.
Species boundary validation using BPP
To further evaluate lineage boundaries surrounding the F. inflexum assemblage, BPP analyses were conducted under two alternative lineage hypotheses representing fine-scale and conservative clustering scenarios, respectively (Supplementary Fig. 7).
Under the 17-lineage scenario, posterior probabilities varied substantially depending on prior settings. When the small ancestral population size and shallow divergence prior combination [θ ~ G(2, 2000), τ0 ~ G(2, 2000)] was applied, both A10 and A11 analyses supported the 17-lineage scenario, with posterior probabilities of 0.96-0.97 and 0.81-0.82, respectively. In contrast, under the two remaining prior combinations (the large ancestral population size and deep divergence prior [θ ~ G(1, 10), τ0 ~ G(1, 10)] and the large ancestral population size and shallow divergence prior [θ ~ G(1, 10), τ0 ~ G(2, 2000)]), the 15- and 16-lineage scenarios received the highest posterior probabilities in the A10 and A11 analyses, respectively. However, support for both scenarios remained weak, with posterior probabilities of only 0.26-0.29 for the 15-lineage scenario (A10) and 0.33-0.34 for the 16-lineage scenario (A11) (Table 2).
In contrast, the conservative five-lineage scenario received consistently strong support across all analyses and prior combinations. Under the A10 model, all runs recovered the five-lineage hypothesis with maximal posterior probability (PP = 1.00). Similarly, A11 analyses strongly supported the same scenario, with posterior probabilities ranging from 0.90 to 0.99 (Table 2).
Diversity of FOSC species and their host/substrate associations in Korea
Based on the current taxonomic framework adopted herein, 11 phylogenetic species were recognized among 92 FOSC isolates from Korea, including F. contaminatum (2 isolates), F. cugenangense (40), F. curvatum (6), F. duoseptatum (2), F. inflexum (17), F. languescens (4), F. nirenbergiae (12), F. odoratissimum (1), F. oxysporum sensu stricto (s. str.) (1), F. triseptatum (4), and F. wimaladesilvae (3) (Table 1). Among these, F. contaminatum, F. curvatum, F. duoseptatum, F. inflexum, F. languescens, F. nirenbergiae, F. odoratissimum, F. triseptatum, and F. wimaladesilvae are newly recorded from Korea. A review of published literature revealed that most previous reports identified isolates within the FOSC only as F. oxysporum or assigned them to formae speciales. Taken together, published records and the results of the present study indicate that only 11 FOSC species have been documented in Korea to date.
Among the phylogenetic species recognized in this study, F. cugenangense was associated with the largest number of plant species, including members of the Amaryllidaceae, Apiaceae, Araliaceae, Asteraceae, Brassicaceae, Convolvulaceae, Cucurbitaceae, Fabaceae, Lamiaceae, Liliaceae, Pedaliaceae, Poaceae, Polygonaceae, Rosaceae, and Solanaceae. Fusarium curvatum was isolated from orchids (Cymbidium kanran, Cymbidium sp., and Phalaenopsis sp.), lettuce (Lactuca sativa), and stock (Matthiola incana). Fusarium nirenbergiae was associated with carnation (Dianthus caryophyllus), gladiolus (Gladiolus × gandavensis), sweet potato (Ipomoea batatas), and tomato (Solanum lycopersicum). In contrast, most of the remaining phylogenetic species were recovered from only one or a few plant species (Supplementary Table 2).
According to published literature and the findings of the present study, members of the FOSC have been recorded from 91 plant species and human clinical specimens in Korea. These plant species represent a broad range of categories, including vegetables, legumes, cereals, medicinal plants, ornamental plants, fruit crops, and forest trees. Among these, many economically important crops in Korea have been reported to be affected by diseases caused by FOSC species, including cabbage (Brassica spp.), pepper (Capsicum annuum), onion (Allium cepa), garlic (Allium sativum), soybean (Glycine max), radish (Raphanus raphanistrum subsp. sativus), cucumber (Cucumis sativus), carrot (Daucus carota), and strawberry (Fragaria × ananassa). However, in many previous studies, fungal species identification was based solely on morphological characteristics or limited molecular data, making the resulting species identifications uncertain.
Discussion
The FOSC remains one of the most taxonomically challenging groups within Fusarium due to its extensive genetic diversity, evidence of recombination, and often ambiguous species boundaries among closely related lineages (Achari et al., 2020; Laurence et al., 2014; McTaggart et al., 2021). Based primarily on multilocus phylogenetic evidence, the present study supports the recognition of 32 phylogenetic species among 49 previously described taxa. Although GCPSR and multispecies coalescent-based analyses supported the distinctiveness of several well-supported species, they also revealed broader evolutionary groupings encompassing multiple closely related taxa. These findings indicate that some currently recognized species may represent relatively recent divergences with limited genealogical differentiation. Nevertheless, the available evidence does not yet justify broader taxonomic revisions, and the 32-species framework is considered the most appropriate interpretation of species boundaries based on current data. The remaining areas of uncertainty highlight the need for additional evidence from genomic, ecological, and biological sources to further refine species delimitation within the FOSC.
Fusarium oxysporum was originally described by Schlechtendal (1824) and was later designated with an epitype by Lombard et al. (2019), providing a modern taxonomic reference for the species complex. Since this revision, the number of recognized phylogenetic species within the FOSC has increased rapidly, reaching up to approximately 49 species according to the updated FUSARIOID-ID database and recent taxonomic studies (Han et al., 2025; Zhao et al., 2025). This expansion has greatly improved our understanding of evolutionary diversity within the complex, but it has also raised questions regarding the status of some recently described taxa. Many FOSC species exhibit limited sequence divergence across commonly used loci, and diagnostic morphological characters remain scarce or overlapping among taxa (Han et al., 2025; Lombard et al., 2019; Zhao et al., 2025). Consequently, distinguishing independently evolving species from structured populations remains a major challenge in FOSC taxonomy. This difficulty likely reflects the complex evolutionary history of the group. In recently diverged lineages, incomplete lineage sorting may generate phylogenetic patterns that resemble species-level divergence despite limited evolutionary independence (Degnan and Rosenberg, 2009). Furthermore, low sequence divergence, recombination, and potential gene flow can produce ambiguous genealogical signals among loci (Fourie et al., 2009; Laurence et al., 2012; McTaggart et al., 2021). These evolutionary processes may obscure species boundaries in some portions of the FOSC and help explain why additional evidence from population-level and genome-scale datasets is often required to confidently resolve relationships among closely related lineages.
Previous studies have conclusively resolved some taxa as synonyms, including F. callistephi, F. purpurascens, and F. tardicrescens, which have been synonymized with F. inflexum, F. odoratissimum, and F. fabacearum, respectively (Crous et al., 2021b; Han et al., 2025; Lombard et al., 2022). In contrast, several other taxa have remained only tentatively proposed as synonyms due to insufficient data and limited analytical frameworks, including F. coriorum as a synonym of F. curvatum; F. cili, F. phialophorum, and F. rosae-roxburghii as synonyms of F. odoratissimum; and F. houttuyniae and F. liriopes as synonyms of F. nirenbergiae (Han et al., 2025; Zhao et al., 2025). In the present study, these previously proposed synonym relationships were consistently supported by multilocus phylogenetic analyses, GCPSR, and STACEY, providing additional evidence for their proposed synonymy. Furthermore, the present analyses recovered several additional pairs or groups of taxa as phylogenetically indistinguishable based on the available loci. These relationships suggest potential synonymy, including F. liaoningense and F. vaughaniae as potential synonyms of F. cugenangense; F. hordeum as a potential synonym of F. agrestense; F. fabacearum as a potential synonym of F. inflexum; F. sangayamense as a potential synonym of F. kalimantanense; F. moutoniae, F. pharetrum, and F. veterinarium as potential synonyms of F. contaminatum; and F. idabrowneae and F. landiae as potential synonyms of F. nirenbergiae. However, given the low sequence divergence and incongruence among loci within the FOSC (McTaggart et al., 2021), these synonym hypotheses require further evaluation using additional molecular data and complementary evidence, particularly for taxa represented by limited sequence data from ex-type strains.
Although multiple analytical approaches in this study also suggested the possibility of broader synonymization in some lineages (F. inflexum, F. languescens, and F. nashariae lineages), these hypotheses will require further evaluation using denser population-level sampling, genome-scale datasets, and additional biological evidence capable of providing substantially greater evolutionary resolution than currently available multilocus markers. The F. inflexum clade represents a notable example of this uncertainty. Multilocus phylogenetic analyses, GCPSR, STACEY, and BPP collectively suggested that many of the recognized species within this clade may represent either intraspecific structure or recently diverged evolutionary lineages that remain difficult to distinguish using currently available loci.
The 32-species framework and the 18-lineage scenario reflect the inherent complexity of species delimitation in the FOSC rather than representing conflicting taxonomic hypotheses. Within the 18-lineage scenario, each lineage corresponds to a distinct clade in the multilocus phylogeny and may comprise either a single phylogenetic species or multiple closely related species. Moreover, several phylogenetic species with uncertain boundaries were consistently supported by both the GCPSR and STACEY analyses. Among the alternative species delimitation scenarios inferred by STACEY, the 18-lineage scenario was proposed because it showed the greatest overall concordance with the multilocus phylogeny and the species hypotheses inferred from both the GCPSR and BPP analyses. Accordingly, the 18 lineages should be regarded as putative evolutionary lineages inferred from the currently available data rather than as formally recognized species. They are intended to illustrate the degree of genetic relatedness among the phylogenetic species recognized in this study and to provide a working evolutionary framework that can be further tested and refined as additional genomic, phenotypic, ecological, and reproductive evidence becomes available.
The inclusion of Korean isolates not only revealed additional phylogenetic diversity within the FOSC, but also provided new insights into the diversity and host/substrate association of FOSC species in Korea. Historically, most Korean isolates were identified as F. oxysporum or assigned to various formae speciales based on morphology, pathogenicity, or limited molecular evidence (Choi et al., 2015; Jeon et al., 2018; Jeon et al., 2017; Joa et al., 2018; Kim et al., 2005; Kwon et al., 1998; Lee et al., 2023; Nam et al., 2017; Park et al., 2019). Re-identification using the analytical frameworks employed in this study revealed substantially greater species diversity than previously recognized. Eleven phylogenetic species were identified among Korean isolates, nine of which represent new records for Korea. Particularly, only a single isolate examined in this study was assigned to F. oxysporum s. str., whereas all remaining isolates belonged to other phylogenetic species within the complex. Among the identified species, F. cugenangense was the most commonly isolated species, followed by F. inflexum and F. nirenbergiae. The results also provide additional insights into the relationship between phylogenetic species and forma specialis designations. Many isolates previously regarded as different F. oxysporum formae speciales were reassigned to the same phylogenetic species, whereas isolates assigned to the same forma specialis were frequently distributed among multiple phylogenetic species (Table 1). Taken together, these findings indicate that reliance on traditional morphology-based identification and forma specialis classifications has obscured the true diversity of the FOSC in Korea, whereas integrative approaches based on molecular data provide a more accurate framework for species identification and classification. Such taxonomic reassessments are essential for improving our understanding of species diversity and host/substrate associations in the country, thereby contributing to more effective disease management strategies and plant quarantine programs.
In conclusion, the present study provides a comprehensive evaluation of species relationships within the FOSC using multiple complementary analytical approaches. Although 32 phylogenetic species were recognized among 49 previously described taxa, the proposed synonymy of several taxa requires further evaluation using additional lines of evidence. Furthermore, the overall concordance among multilocus phylogenetic analyses, GCPSR, and multispecies coalescent-based approaches supports a tentative hypothesis of 18 evolutionary lineages within the FOSC. Nevertheless, broader taxonomic revisions should await further investigation based on denser population-level sampling, genome-scale datasets, and additional biological evidence. The study also provides a comprehensive assessment of numerous Korean FOSC isolates, revealing previously unrecognized diversity and substantially expanding our knowledge of the host and substrate associations of FOSC species in Korea.
PDF Links
PubReader
ePub Link
Full text via DOI
Download Citation
Supplement1
Print
