Metabarcoding pollen to identify allergenic species of grasses (Poaceae) and ragweed (Ambrosia spp.) across six monitored sites in South Africa

Sidla, Siyavuya

Metabarcoding pollen to identify allergenic species of grasses (Poaceae) and ragweed (Ambrosia spp.) across six monitored sites in South Africa

Thesis / Dissertation

2025

Publisher

University of Cape Town

Department

Allergology and Clinical Immunology Unit

Faculty

Faculty of Health Sciences

Abstract

Background and aims Aeroallergens contribute to the global burden of non-communicable diseases, causing allergic conditions such as rhinitis, conjunctivitis, and asthma. Pollen is the second most important contributor, and despite cross-reactivity between allergens, patients often experience distinct sensitivity patterns to different species. The occurrence of allergenic species varies with geography, seasonality, and their flowering periods are influenced by meteorological factors and climate. The South African Pollen Network (SAPNET) monitors aerospora in different biomes of the country, providing weekly reports on the prevalent species and their concentrations, raising awareness among allergy sufferers and healthcare providers. Grass pollen is one of the most abundant contributors to airborne allergies. However, current SAPNET methods of analysing pollen, using light microscopy, can only identify grasses up to the family level and ragweed (Ambrosia spp.) pollen to the genus level. This project aimed to introduce DNA metabarcoding techniques to classify and identify grass and ragweed species contributing to pollen allergies in various South African biomes. Methods Samples were selected from weeks with the highest grass pollen counts between October 2022April 2023 and included samples from sites where Ambrosia pollen was identified (Durban and Potchefstroom). DNA was isolated from environmental pollen samples using the NucleoMag bead beating method, optimised to obtain a larger quantity of double-stranded DNA (dsDNA) and further purified with an additional OneStep PCR inhibitor removal step to improve purity. DNA metabarcoding analysis was performed using two taxonomic markers: ribulose-1,5bisphosphate carboxylase (rbcL) and internal transcribed spacer 2 (ITS2), amplified with universal primer pairs in a two-step polymerase chain reaction (PCR) library preparation for Next Generation Sequencing (NGS) with MiSeq Illumina V2 systems. The sequence reads generated were assessed for quality, pre-processed, and assigned to taxa using two bioinformatics pipelines. The first followed the method for analysing metabarcoding dualindex data described by Sickel et al. (2015), while the second was adapted from the National Botanic Garden of Wales Plant Illumina pipeline developed by Ford and Jones (2021). Results and Discussion The DNA extraction protocol was optimised by altering the bead beating method using the samples from weeks with high grass pollen concentrations. From the improved method of extraction, we were able to obtain sufficient DNA yield with an average of 21.88 ng/µL. Additional purification improved the dsDNA A280/260 purity ratio to within the ideal range for 73% of the samples, though it introduced salts that decreased the A230/260 ratio below the optimal range. Amplification of rbcL and ITS2 barcodes initially produced amplicons of 579 bp and 542 bp, respectively. However, to limit sequencing costs, an alternative set of ITS2 primers producing amplicons were explored. A total of 25 samples progressed to NGS: Cape Town (7), Kimberley (6), Bloemfontein (3), Johannesburg (3), Durban (3), and Potchefstroom (2), and six ITS2 samples were excluded during quality check due to short amplicon lengths below 100 bp. The three taxonomic classification tools; UTAX, RDP, and BLAST were used to assign taxa, with a 50-read cutoff applied as the minimum read count, with confidence scores of ≥80% for UTAX and ≥0.8 for RDP. For BLAST classifications, only sequences with 99 - 100% identity was considered, using thresholds of ≥80% query coverage and an E-value of ≤1.00×10-5. UTAX classified rbcL and ITS2 sequences identified 233 unique Poaceae species, including Lolium perenne (ryegrass), which was one of the 17 species identified by both barcodes. RDP classifications resulted in 48 Poaceae species, with Poa annua (annual bluegrass) and Cynodon dactylon (Bermuda grass) being two of the three species classified by both barcodes. BLAST classifications produced the largest biodiversity, identifying 307 unique Poaceae species with 19 species common to both barcodes including Cynodon dactylon and Lolium perenne. Additionally, Paspalum notatum (Bahia grass), Phleum pratense (Timothy grass), Phragmites australis (common reed) and Stenotaphrum secundatum (buffalo grass). Two ragweed species, Ambrosia artemisiifolia and A. trifida were positively identified from rbcL sequences with classifications using the three tools. Conclusions This pilot study successfully used pollen metabarcoding to identify known allergenic grasses in several South African regions, including Bermuda, rye, Timothy, common reed and annual bluegrass in Cape Town; Bermuda, rye, and annual bluegrass in Kimberley; Bahia and Bermuda grass in Johannesburg; weeping love and Bermuda grass in Bloemfontein; and Ambrosia artemisiifolia and A. trifida in Durban. The short length of DNA libraries limited sequence read length, which significantly influenced the classification of operational taxonomic units (OTUs). The low confidence in taxa assignments from UTAX classifications and the identification of species via BLAST that are not recorded in South Africa raised concerns regarding the reliability of UTAX as a taxonomic classification tool and the NCBI based reference databases, which are primarily developed based on Northern Hemisphere plant data and may misclassify endemic species. Future research should address these limitations and include a larger number of SAPNET samples to better map the seasonality and distribution of Poaceae species across South African biomes.