Introduction: Pollen metabarcoding represents a valuable tool for the characterisation of pollen loads and pollination networks in agricultural landscapes. However, a significant limitation lies in the frequent lack of comprehensive taxon coverage within plant reference databases. This is especially the case for the Afrotropical Region. With the general objective of characterising the pollination networks in a large experimental setup established in Morogoro (East-Central TZ), we assembled a library of reference DNA barcodes from plants from this area. Methods: Herbarium vouchers were collected in plots where agroecological and conventional cucurbit farming was implemented. Duplicated herbarium samples were obtained from the same plant, morphologically identified and deposited at the Sokoine University of Agriculture in Tanzania and the herbarium of Meise Botanic Garden in Belgium (herbarium code: BR). DNA was extracted from leave clips and used to amplify four loci (matK, rbcL, ITS1, ITS2). Results: The 140 samples collected belong to 106 plant taxa from 33 families. The plant DNA barcodes obtained for the four target markers have been made accessible through the Barcoding of Life Data Systems (BOLD, project: ISeBAF). Conclusions: These data will further improve the performance of metabarcoding identification in pollen loads from Afrotropical pollinators.