Isolates used in this study were collected from multiple locations across Vietnam in 2018, as summarized in Table 1 and displayed geographically in Supplementary Figure 1. Elevation, substrate, prevalent plants, province, and geographic coordinates were recorded and maintained as part of the dataset.

Two methods were used to isolate oomycetes based on the type of the sample taken. Soil samples (10 cc) were collected in disposable cups and brought to the laboratory, where they were flooded with sterile water and kept at 18–20 °C under natural light. Pieces of leaflets aged 3–10 days from Lithocarpus bacgiangensis, Quercus glauca, Q. gilva, Castanopsis indica, Chamaecyparis hodginsii, or Acacia mangium were placed on the flooded soil, where they served as baits for newly emerging zoospores. After 2–3 days, a subset of the leaflets developed necrotic lesions characteristic of Phytophthora infection. Lesion margins were surface-disinfected, air-dried, and plated onto PARPNH agar, as described by Erwin and Ribeiro (1996). PARPNH consists of 20% V8 agar containing 10 mg pimaricin, 200 mg ampicillin, 10 mg rifampicin, 200 mg pentachloronitrobenzene, 50 mg nystatin, and 50 mg hymexazol per liter (Jung et al., 2000). PARPNH is semi-selective for oomycetes. For river samples, water was collected, and leaves of Castanopsis sp. or Quercus sp. were placed on the surface of the water to serve as baits. Following the development of necrotic lesions, the organisms were isolated and processed as described above.

Resulting colonies were subcultured by transferring small fragments of mycelia to obtain organisms in pure culture. Isolates were stored at 4 °C on the colonized plates of PARPNH agar and annually subcultured to maintain viability.

To extract genomic DNA, isolates were grown in pure culture on clarified 20% V8 juice agar for 1 week. Aerial mycelium and the upper layer of the media plate were gently scraped with a sterile scalpel, and genomic DNA was extracted using the FastDNA soil extraction kit (FastDNA, MP Biomedicals, Solon, OH). The genomic DNA was used as a template in standard PCR reactions using internal transcribed spacer (ITS) 6/4 (White et al., 1990; Cooke et al., 2000) and cox2 primers (Hudspeth et al., 2000). PCR products were visualized by gel electrophoresis to confirm amplification based on the presence of 1 kb and 640 bp bands for ITS and cox2, respectively.

PCR amplicons were purified using ExoSap-IT (Applied Biosystems; Thermo Fisher Scientific, Carlsbad, CA, USA) and submitted for Sanger sequencing (Eurofins Inc., Louisville, KY, USA). Initial reactions were sequenced with the ITS6 primer, but the samples that yielded low-quality sequences were additionally sequenced in the reverse direction using the ITS4 primer. In these cases, forward and reverse sequences were aligned, edited, and analyzed using BLASTn. To confirm or clarify identification, amplified fragments of the cox2 region were bidirectionally sequenced, aligned, edited, and analyzed with BLASTn.

To determine the diversity of oomycetes detected in this study, phylogenetic analyses were conducted using sequences of the rDNA internal transcribed spacer (ITS) and the cytochrome c oxidase 2 subunit (cox2). The ends of each sanger sequence were trimmed using the Trim End function in Geneious Prime (2023.0.1; Biomatters Inc., Boston, MA, USA) with the corresponding primer pair. When available, a consensus sequence was generated from each bidirectional sequence for both ITS and cox2 using the De Novo Assemble function in Geneious Prime. Representative sequences of known species were retrieved from GenBank for Phytopythium, Phytophthora, Pythium, and the outgroup Globisporangium paroecandrum (Supplementary Table 1). The representative sequences were aligned with the Vietnam isolate sequences using MAFFT (v7.490) (Katoh and Standley, 2013), and terminal regions, including primer regions, were trimmed using Geneious Prime. Sequences with missing ends were replaced with “?” to denote missing data. The resulting alignment was analyzed with IQ-TREE (v2.2.3) (Nguyen et al., 2015) to generate maximum likelihood trees with automatic model selection (Kalyaanamoorthy et al., 2017). The resulting Newick tree files were visualized using FigTree v1.4.4.¹ The tree was rooted using the consensus sequence of Globisporangium paroecandrum CBS 157.64 (Figure 1). This process was done to generate separate ITS and cox2 gene trees. A concatenated tree was not generated because most reference Phytophthora and Pythium sequences do not have cox2 sequences.

Sequences were deposited in NCBI GenBank: ITS accessions OR531129–OR531278 and cox2 accessions OR704574–OR704733. Isolates showing >99% sequence identity to a deposited reference sequence were classified as that species, provided that these groups were supported by phylogenetic analysis.

Over 700 isolates were obtained from the 93 sites sampled in 2018; however, less than half were revived following long-term storage. Sequences were generated from 245 isolates, corresponding to 21 unique oomycete species. The majority of isolates (66.5%) were identified as Phytopythium spp., followed by Phytophthora spp. (31%) (Table 2) and Pythium spp. (2.4%). Phytopythium spp. accounted for 93.3% (28 of 30) of isolates obtained from agricultural soils, 65.6% (118 of 180) from forest soils, and 48.6% (17 of 35) from river samples. Oomycete taxa obtained from the different types of sites (agricultural soil, forest soil, and water) are summarized in a heatmap (Figure 2).

Phytopythium vexans and Phytophthora cinnamomi were the most frequently isolated species, accounting for 77.9 and 78.9% of isolates within their respective genera (Figure 2). Just over 6.9% (17 out of 245) of sequenced isolates could not be assigned to a species due to either having high (>98%) nucleotide similarity to multiple GenBank entries purportedly representing different species or not sharing high similarity with any entries. Among Phytopythium taxa, Pp. vexans, which was recovered from all three site types, was the most frequently isolated species (77.9%), followed by Pp. helicoides (7.6%) and Pp. iriomotense (5.7%). Species-level identification could not be made for nine Phytopythium isolates originating from citrus soil and river water, and they are labelled here as Phytopythium sp. 1 (n = 5) and Phytopythium sp. 2 (n = 3) (Table 2).

Phytopythium sp. 2 shared 99% identity with sequences of undescribed Phytopythium species previously reported by Jung et al. (2020) in rivers in Vietnam (MN872739) and with another isolate obtained from forest soil in Poland (KC602492). BLASTn analysis of the cox2 sequences showed the closest match to Pp. helicoides (MN952211); however, they only shared 96.7% nucleotide identity, which is below the threshold expected for a member of the same species, based on other isolates in this study. A total of 11 Phytophthora taxa were obtained, with five represented by a single isolate.

Among the isolated Phytophthora spp., most isolates (n = 60; 78.9%) belonged to P. cinnamomi, followed by P. lagoariana (n = 3) and P. pini (n = 3). The majority of Phytophthora isolates were obtained from forest soils (81.6%), with only 17.1 and 1.3% from the river and agricultural soil samples, respectively. A total of six isolates appeared to belong to three undescribed species of Phytophthora, with the ITS sequence of P. sp. 1 grouping equally closely with P. pini and P. citricola (Figure 1). P. sp. 2 shared the highest nucleotide identity (92.3%) with a sequence of P. plurivora (KT693124) from forest soil in Hungary, followed by 92.2% shared identity with a sequence of P. polonica (KT693123). P. sp. 3 showed greater than 99% sequence identity with numerous taxa, including two putative hybrids: P. sp. zentmyerii × Peru4-like from Portugal (PQ238050) and P. sp. × Kunnunara-like from Taiwan (KU682602).

Pythium amaminum (n = 2) and Py. dissotocum (n = 1) were isolated from river water, while a third species, Py. sp. (n = 3), was isolated from both river water and citrus soil. This unidentified species could not be confirmed based on GenBank matches and may represent an undescribed species. In the phylogenetic analysis, it clustered closely with the reference isolate Py. torulosum (CBS 166.68_AY598634), despite 23 mismatches and 14 gaps between the two sequences (Figure 1). The three isolates of the unknown Pythium species shared ≥99.6% identity in cox2 sequences and ≥99.3% identity in ITS sequences. These high levels of sequence similarity justified their classification within the same species.

The ITS alignment included 72 sequences (32 from this study) with 1,309 columns, 1,007 distinct patterns, 666 parsimony-informative sites, 147 singleton sites, and 495 constant sites. The best-fit model was TIM3 + F + I + G4. The ITS sequences of Phytopythium, Phytophthora, and Pythium spp. obtained in this study grouped with well-established reference isolates (Figure 1A). The two new Phytopythium species grouped with members of clades 2, such as Pp. fagopyri and Pp. palingenes. Among the undescribed Phytophthora species, P. sp. 1 clustered within clade 2, while P. sp. 2 and 3 fell within clade 9. The single undescribed Pythium species isolated grouped with Py. torulosum, which is part of clade B of this genus.

The cox2 alignment contained 47 sequences with 594 columns, 248 distinct patterns, 180 parsimony-informative sites, 58 singleton sites, and 356 constant sites. The best-fit model was TPM2u + F + I + R3. Few reference isolates had cox2 sequences available, so these were not included in the second phylogenetic analysis (Figure 1B). Despite this, the cox2 and ITS trees shared similar architecture (Figure 1).