From 2018 to 2021, Orchidaceae plants, including Calanthe speciose (Cas), Cattleya (Cat), Chysis limminghis (Ch), Cymbidium (Cy), Dendrobium (De), Epidendrum (Ep), Haraella retrocalla (Ha), Maxillaria (Ma), Oncidium (On), Paphiopedilum (Pa), Phalaenopsis (Ph), Renanthera (Re), Tuberolabium kotoense (Tu), Vanda (Va), and Vanilla planifolia (Vap) showing rot or blight in roots, stems, leaves, or shoots, were collected from orchid gardens located in Changhua, Chiayi, Nantou, Pingtung, Taichung, Tainan, Taoyuan, and Yunlin in Taiwan. Eight locations were visited, with 33 investigations. Pieces of diseased tissues (5 × 5 mm) were cut and sterilized with 1% NaClO for 30 s, followed by rinsing three times with sterilized distilled water, dried using sterilized paper, and cultured on water agar (2% WA; Fei Kung Agar-Agar, Tainan, Taiwan) at 28°C with 12-h light daily for 1–3 d. Then, a single mycelium was cut and cultured on potato dextrose agar plates (PDA; BD Difco™, New Jersey, USA) at 28 °C with 12-h light daily for 5–7 d. The morphologies of colonies and conidia were observed according to the methods described by Leslie and Summerell (2006). FOSC-like isolates were cultured in pure culture medium using single spores. The purified isolates were cultured on PDA at 28°C with 12-h light daily for 7 d; then, two 2–3 mm² agar plugs from the colony margins were transferred to 5-mL glass store tubes (with 10% sand soil and 1% agar) and maintained for long-term storage.

The FOSC isolates used in this study satisfied Koch’s postulates. The plants used in pathogenicity test were recorded in Table 1. Most isolates were inoculated on the same species from which they were originally obtained. However, some isolates were tested on different species because of limitations in price or availability. The orchids used for inoculation were 1–2 years old plants. Some of them inoculated with tissue culture seedlings were 4–6 months, such as Ph. The FOSC isolates were grown on PDA plates for 7–14 d at 28°C with a 12-h light period per day. The orchid plants were sterilized using a paper towel with 75% ethanol, and wounds were created using a needle (1–5 mm depth) on the leaves, stems, pseudobulbs, or crowns of the orchid plants. Inoculation was conducted by two methods, mycelium plug inoculation and spore suspension inoculation. The mycelium inoculation method was adapted from Chung et al. (2011). Three mycelium plugs were placed on each wound site, and PDA agar plugs were used as the control treatment. The spore suspension inoculation method was adapted from Huang et al. (2014). The spore suspension (1 × 10⁷ spores/mL) was prepared by washing spores with sterilized ddH₂O, followed by filtration through Miracloth. The 10–20 μL spore suspension mixed 1:1 (v/v) with 0.2% WA was dropped on each of the four wounds, and sterilized ddH₂O was used as negative control. The inoculation methods for each isolate were shown in Table 1. Inoculated plants were incubated in a growth chamber at 28°C under a 12-h light period per day, and symptoms were recorded 7–14 d after inoculation, depending on the orchid species.

DNA extraction was performed as described by Saitoh et al. (2006) with some modifications. After culturing on PDA at 28°C with 12-h light daily for 10 d, the mycelia of isolates were scraped and transferred into 1.5-mL microcentrifuge tubes with 500 μL of a lysis solution (200 mM Tris-HCl, 50 mM ethylenediaminetetraacetic acid, 200 mM NaCl, and 1% n-lauroylsarcosine sodium salt at pH 8.0) and placed at −20°C for 24 h. Once lysis was completed, 500 μL of phenol:chloroform:isoamyl alcohol (25:24:1) was added to the tubes and mixed gently. The mixture was then centrifuged at 16,200 × g for 10 min. Next, the supernatant was collected in a new microcentrifuge tube with 240 μL of isopropanol and placed at −20°C for 1 h. To obtain the DNA pellet, the mixture was centrifuged at 16,200 × g for 5 min, and the supernatant removed, followed by the addition of 700 μL of 70% ethanol at −20°C, centrifuged at 16,200 × g for 1 min, and the supernatant removed. After placing the DNA pellet in a laminar flow hood to dry, 30 μL milli-Q water was added, and the mixture was placed at 5°C in an incubator for 15 min. The DNA samples were stored at −20°C for subsequent analysis.

The primer pairs and PCR program used in this study are listed in Supplementary Table S1. For FOSC amplification, cmdA, rpb2, tef1, and tub2 were used. Primers Cal228F/CAL2Rd were used for cmdA (Carbone and Kohn, 1999), RPB2-F/RPB2-R for rpb2 (Wu et al., 2023), EF1/EF2 for tef1 (O’Donnell et al., 1998), and T1/CYLTUB1R for tub2 (O'Donnell and Cigelnik, 1997; Crous et al., 2004). According to Lombard et al. (2019), tef1 provides the highest resolution among these gene regions. Therefore, tef1 was selected for preliminary identification of isolates. The PCR mixtures contained 5 μL of PCR Master mix II (Genemark Technology Co., Ltd, Taiwan), 0.5 μL of each 10 mM primer, 18 μL of milli-Q water, and 1 μL of the DNA sample. All PCR products were subjected to electrophoresis analysis on a 1% TAE (40 mM Tris, 20 mM sodium acetate, and 1 mM ethylenediaminetetraacetic acid, pH 7.5) agarose gel to check their size and were sent to Tri-I Biotechnology Co., Ltd., Taiwan for purification and sequencing. To ensure sequence accuracy, PCR products were subjected to bidirectional sequencing and were assembled using BioEdit version 7.0.5.3 (Hall, 1999). Chromatograms were carefully reviewed to confirm base calling. The sequences were uploaded to National Center for Biotechnology Information (NCBI) via DDBJ.

The assembled DNA fragments were aligned with reference sequences (Supplementary Table S2) from the NCBI database using MEGA version 7.0.26 (Kumar et al., 2016) with ClustalW. The aligned sequence fragments of each gene or region were merged for multilocus phylogenetic analyses. Four loci, cmdA, rpb2, tef1, and tub2, were analyzed. The merged sequences were subjected to Maximum Likelihood analysis to identify related taxa. To confirm the best evolutionary model, JModelTest version 2.0 (Posada, 2008) was used. Maximum likelihood analyses were performed with RAxML-ng v1.2.2 (Kozlov et al., 2019) using 1000 bootstrap replicates. MrBayes v.3.2.6 (Ronquist and Huelsenbeck, 2003) was used to construct a Bayesian inference tree. A Markov Chain Monte Carlo algorithm was used to calculate the random tree topology, which lasted for at least 4.5 M generations. A Markov Chain Monte Carlo analysis was performed until the average standard deviation of the split frequencies was below 0.01 with trees saved every 100 generations. The first 1,000 trees were discarded, and the remaining trees were used to determine the posterior probabilities.

Single conidia of FOSC isolates were cultured on PDA and incubated at 28°C with 12-h light daily for 7 d in the incubator. PDA was used to record colonies and colonial pigments. Carnation leaf-piece agar (CLA) was used to observe FOSC conidial characteristics (Fisher et al., 1982). FOSC isolates were cultured on CLA plates and incubated at 24°C with 12-h blue light daily for 10 d. Conidial characteristics included the shape and size of the microconidia and macroconidia. Spezieller Nährstoffarmer agar was prepared to observe the chlamydospores. The conidial morphology of the isolates was observed using a Zeiss EL-Einsatz Axiophot 156 microscope (Carl Zeiss, Jena, Germany), and images were recorded using a Zeiss Axiocam 105 color camera (Carl Zeiss).

The orchids collected in this study are listed in Table 2, including 63 FOSC isolates. Most were collected from cultivation facilities, whereas others were collected from flower markets throughout Taiwan (Table 1). The source, origin, longitude and latitude, and collection date were recorded in Table 1. Eight major orchids, including Cat, Cy, De, Ha, Ma, Pa, Va, and Vap, were collected (Figure 1). Among the diseased plants, stem/pseudostem/pseudobulb rot was the predominant symptom in most orchids, and lesions on the leaves were minor (Table 2, Figure 1). The most common symptoms were pseudobulb rot in Cy, basal leaf dry rot in Pa, and stem rot in Vap (Table 2, Figures 1B, F, H). In the case of pseudobulb rot in Cy, leaf yellowing may show at the initial stage, then the entire pseudobulb might become completely rotted with foul smell develops (Figure 1B); meanwhile, the diseased Cy dies latterly. Here, we found the pathogens might remain latent until conditions become favorable for disease development. For stem rot in De, certain rot symptoms involve yellowing of the entire stem. In the case of basal leaf dry rot or leaf blight in Pa, the rot begins at the leaf base and gradually expand to approximately half of the leaf, ultimately causing symptomatic leaves to detach from the plant (Figure 1F). For stem and basal leaf rot in Va, rot initially occurs in the stem, and as the disease progresses, pronounced symptoms develop at the bases of young leaves, accompanied by yellowing (Figure 1G). In the case of Vap, leaves, stems, roots and beans have the potential to be infected. The roots show rot with yellowing. They also have latent infections in Vap, which may become symptomatic when Vap is unhealthy or lacks water. Finally, the diseased section of the stem dies off (Figure 1H). Based on the tef1 sequence, these Fusarium-like isolates were confirmed as FOSC. The results showed that FOSC were the dominant pathogens causing diseases in terrestrial orchids (Cy and Vap). Fewer FOSC isolates were obtained from epiphytic orchids. Only a few FOSC isolates were obtained from Cat, Ha, and Ma. Therefore, these three orchids may not be major hosts of FOSC in Taiwan.

In this study, 63 FOSC isolates were examined and their pathogenicity was confirmed using Koch’s postulates (Tables 1, 2). The pathogenicity test showed that all isolates could infect the original hosts or alternative testing hosts (some isolates from De, Pa, and Va), resulting in similar symptoms on the leaves, basal leaves, pseudostems, basal stems, pseudobulbs, or (Table 1). Figure 2 displayed symptoms after inoculation, with one representative image shown for each orchid. Day post-inoculation and inoculation methods for each isolate were specified in the figure legends.

The test included 63 FOSC isolates obtained from orchids in this study, four isolates from Anoectochilus formosanus (Af) (F7, Le91, F1, and FL1409) (Huang et al., 2014; Hsu, 2016), two FOSC isolates from Cy. ensifolium (Fo-92 and Fo-51) (Huang et al., 2020), and three isolates from Phalaenopsis spp. (Ph) (FuC2r, FuTn7s, and FuTn29r) supplied by Dr. Wang (Developmental Biology of Phytopathogenic Fungi Lab, National Chung Hsing University, Taiwan), and one isolate from Ph (N8284) provided by Dr. Su (Plant Pathology Division, Taiwan Agricultural Research Institute, Taiwan) (Su et al., 2012) were analyzed together (Supplementary Table S3). Following the system of Lombard et al. (2019) (Supplementary Table S2), the phylogenetic results of the tef1 single gene can provide an initial classification of FOSC, similar to the results obtained from the multilocus phylogenetic analysis. The 73 FOSC isolates were divided into six taxa (Figure 3), including F. contaminatum (five isolates), F. cugenangense (one isolate), F. curvatum (34 isolates), F. nirenbergiae (22 isolates), F. odoratissimum (two isolates), and F. triseptatum (9 isolates), based on the tef1 gene sequence (Table 3). Isolates grouped with F. contaminatum were obtained from Af (two isolates) and De (three isolates). Fusarium curvatum isolates were the dominant species and were obtained from most orchid species, including Cat, Cy, De, Ha, Pa, Ph, and Va, with the majority being accounted for by Cy (20 isolates). Isolates grouped with F. nirenbergiae were obtained from Cy (one isolate), Maxillaria (Ma) (one isolate), Pa (one isolate), Ph (two isolates), Va (three isolates), and Vap (14 isolates), which had the second-highest orchid species diversity. Isolates classified as F. odoratissimum were isolated only from Af (two isolates). The isolates classified as F. triseptatum were from Cy (eight isolates) and Pa (one isolate). To achieve a more precise classification of these isolates, isolates belonging to different putative Fusarium species were used to amplify three additional gene sequences (cmdA, tub2, and rpb2), and phylogenetic analysis was performed based on the four gene sequences. The multigene alignment length was 2,234 bases (cmdA, 573 bases; tef1, 629 bases; tub2, 430 bases; and rpb2, 602 bases). The Maximum Likelihood tree is shown in Figure 4, and the calculated bootstrap and posterior probability values are shown in the branches. The results indicated that the 73 FOSC isolates used in this study were separated into six taxa, similar to the tef1 gene sequence analysis (Table 3). The GenBank accession number of FOSC isolates from orchid were listed in Supplementary Table S3.

Based on phylogenetic inference, the existing FOSC (six species) were determined by multilocus phylogenetic analysis and their morphological characteristics are described below. In this study, F. contaminatum formed white to bright orange colonies on PDA, and aerial mycelia were abundant (Figure 5A). Microconidia were ellipsoidal or falcate-shaped with 0–1 septa, forming false heads, 5.0-(6.6)-14.8 µm in length and 1.8-(2.5)-3.7 µm in width (Figures 5B–E). Macroconidia were falcate with slightly curved and foot-like cells with 2–5 septa, 26.0-(35.5)-47.1 µm in length and 2.8-(3.8)-4.8 µm in width (Figures 5C–G). Chlamydospores grew singly or in pairs, formed in hyphae and in conidia, mostly globose or subglobose, smooth to rough-walled, 6.0-(9.6)-15.1 µm (Figures 5H, I).

Fusarium cugenangense formed pink to bright purple colonies with less abundant aerial mycelia on PDA, and its color on the reverse side was pink to orange (Figure 6A). Microconidia were ellipsoidal or falcate-shaped with 0–1 septa, forming false heads, 7.4-(14.4)-37.5 µm in length and 2.1-(3.0)-4.4 µm in width (Figures 6B, D, F–H, K). It had the largest microconidia on CLA among the species mentioned in this study. Microconidia with 1-septa were more abundant than the other species mentioned in this study (Figures 6B–K). Macroconidia were falcate with more pronouncedly curved and foot-like cells with 1–5 septa, 22.8-(31.7)-39.0 µm in length and 3.4-(4.0)-4.8 µm in width (Figures 6C–G). The 4-septate macroconidia were barely observable (Figures 6J). Chlamydospores grew singly or in pairs, formed in hyphae and in conidia, mostly globose or subglobose, smooth to rough-walled, 6.6-(10.7)-21.1 µm (Figures 6L, M). It had the largest chlamydospores among the species examined in this study.

Fusarium curvatum was a white to bright pink colony with fewer aerial mycelia than the other species mentioned in this study (Figure 7A). Microconidia were ellipsoidal or falcate-shaped with 0–1 septa, forming false heads, 5.5-(7.5)-11.6 µm in length and 1.8-(2.5)-3.4 µm in width (Figures 7B–C, F). Macroconidia were falcate with slightly curved and foot-like cells with 2–4 septa, 23.7-(30.1)-36.0 µm in length and 2.7-(3.6)-4.4 µm in width (Figures 7D–G). Macroconidia were more curved than those of F. contaminatum but not as curved as described by Lombard et al. (2019). Macroconidia of Fusarium curvatum were easily produced on CLA; however, 4-septate macroconidia were barely observed. Chlamydospores grew singly, formed in hyphae and in conidia, were mostly globose or subglobose, smooth to rough-walled, 6.0-(6.8)-7.8 µm (Figures 7H).

Fusarium nirenbergiae sometimes formed white to bright purple or bright orange colonies (Figures 8A). Although it had fewer aerial mycelia, it also had more aerial mycelia than F. curvatum (Figure 8A). Microconidia were ellipsoidal, kidney-, or falcate-shaped with 0–1 septa, forming false heads, 4.8-(7.2)-10.6 µm in length and 1.9-(2.5)-3.3 µm in width (Figures 8B, C, G). Macroconidia were falcate with slightly curved and foot-like cells with 2–5 septa, 28.2-(36.2)-43.7 µm in length and 2.9-(3.9)-4.7 µm in width (Figures 8D–H). Chlamydospores grew singly or in pairs, formed in hyphae and in conidia, mostly globose or subglobose, smooth to rough-walled, 5.5-(7.7)-8.7 µm (Figures 8I, J).

Fusarium odoratissimum had white to bright orange colonies, but the aerial mycelia were less abundant than those of F. contaminatum (Figure 9A). Microconidia were ellipsoidal or kidney-shaped with 0–1 septa, forming false heads, 5.2-(7.9)-10.8 µm in length and 2.1-(2.8)-3.8 µm in width (Figures 9B–G). Macroconidia were falcate, slightly curved, wider in width, and had foot-like cells, with 2–4 septa, 27.5-(34.2)-43.2 µm in length and 3.6-(4.2)-5.1 µm in width (Figures 9B–G). In the present study, the conidia of F. odoratissimum isolates did not possess many septa, as reported by Maryani et al. (2019). Chlamydospores grew singly or in pairs, formed in hyphae and in conidia, mostly globose or subglobose, smooth to rough-walled, 5.8-(7.1)-8.9 µm (Figures 9H, I).

Fusarium triseptatum was purple, with the most abundant aerial mycelia on PDA (Figure 10A). Microconidia were ellipsoidal or falcate-shaped with 0–1 septa, forming false heads, 5.2-(7.5)-10.5 µm in length and 1.8-(2.9)-4.3 µm in width (Figures 10B, F). Macroconidia were falcate with slightly curved and foot-like cells with 2–6 septa, 20.2-(34.4)-52.6 µm in length and 3.1-(4.3)-5.1 µm in width (Figures 10C–G). Among the six FOSC taxa in this study, F. triseptatum had the largest and most abundant septa macroconidia on CLA. Chlamydospores grew singly or in pairs, formed in hyphae and in conidia, mostly globose or subglobose, smooth to rough-walled, 6.5-(8.2)-9.7 µm (Figures 10H, I).