Metabarcoding with Illumina and Oxford Nanopore Technologies provides complementary insights into tree seed mycobiota

Culturing of fungi is labor-intensive and reveals limited diversity, while high-throughput sequencing of barcodes (i.e., metabarcoding) enables a simultaneous detection of fungi from multiple environmental samples. Metabarcoding using short-read sequencers, such as Illumina platforms, provides high sequencing depths but results in many unidentified taxa. Long-read sequencing can improve species and genus assignments but might encompass lower sequencing depth and limit diversity coverage. In this study, fungi in seeds of eleven angiosperm and gymnosperm tree species were assessed using traditional culturing, Illumina short-read metabarcoding, and Oxford Nanopore Technologies long-read metabarcoding. We focused on seed-borne fungi as understanding their diversity and potential impacts on seedlings is crucial for securing plant health. We compared (1) the number and identity of fungal genera and species between metabarcoding approaches and traditional culturing and (2) fungal alpha- and beta-diversity between metabarcoding methods, considering different hosts and fungal lifestyles.

In both short- and long-read metabarcoding datasets, similar numbers of fungal reads and operational taxonomic units were assigned to comparable numbers of fungal genera and species. About one-third of the identified genera were plant pathogens, followed by saprotrophs and endophytes. Culturing overall revealed fewer fungal genera, while most of the fungal reads in short-read metabarcoding datasets stemmed from cultured taxa. Long-read metabarcoding revealed lower per-sample diversity than short-read metabarcoding and distinct fungal communities compared to those from the short-read datasets. Host-dependent patterns in alpha- and beta-diversity were observed across methods, with angiosperms harboring more fungal taxa than gymnosperms, and distinct community structuring across host tree groups and species, although the differences were stronger in short-read than long-read metabarcoding datasets.

Illumina and Oxford Nanopore Technologies metabarcoding captured similar host-dependent diversity patterns despite observed differences in numbers and composition of fungi. Short-read metabarcoding might be optimal for fungal biodiversity studies due to higher sequencing depths and resultant breadth of diversity. As error rates are continuing to decrease, reference databases expand, and throughput improves, long-read metabarcoding is becoming a strong candidate for future diagnostic studies of fungi. Traditional culturing captures most of the fungi from short-read metabarcoding and remains valuable for obtaining isolates for further research.