Globally, biodiversity is at risk of extinction, and megadiverse countries become key targets for conservation. South Africa, the only country hosting three biodiversity hotspots, harbours tremendous diversity of at-risk species deserving to be protected. However, the lengthy risk assessment process and the lack of required data to complete assessments is a serious limitation to conservation since several species may slide into extinction while awaiting risk assessment. Here, we employed deep neural network model integrating species climatic and geographic features to predict the conservation status of 116 unassessed plant species. Our analysis involved in total 1072 plant species and 112 066 occurrence points. The best-performing model exhibits high accuracy, reaching up to 83.6% at the binary classification and 56.8% at the detailed classification. Our best-performing model predicts that 32% and 8% of Data Deficient and Not-Evaluated species are likely threatened, respectively, amounting to a proportion of 24.1% of unassessed species facing a risk of extinction. Interestingly, all unassessed species predicted to be threatened are in protected areas, revealing the effectiveness of the South Africa’s network of protected areas in conservation, although these likely threatened species are more abundant outside protected areas. Considering the limitation in assessing only species with available data, there remains a possibility of a higher proportion of unassessed species being imperiled.

The genus Encephalartos is entirely endemic to Africa. Like most cycad species, the genus is at risk of extinction. One of the threats jeopardizing the future of the genus is reproduction failure. Our objective is to investigate what predisposes Encephalartos species to reproduction failure. We collected functional traits of 430 individuals of Encephalartos villosus, including pre-dispersal seed predation and habitat types and elevation in Origi Gorge Nature Reserve in South Africa. Then, we analysed our data by fitting a structural equation model (SEM). Surprisingly, elevation does not predict pre-dispersal seed predations, adding to the inconsistent effects of elevation on seed predation. However, there was evidence of more predated seeds on plants with more leaves, fitting the patterns of resource concentration hypothesis which predicts more insect herbivores, particularly specialist herbivores, where food resources (e.g., leaves, flowers, fruits, and seeds) are abundant. We also found that more predated seeds are in open habitats, perhaps mirroring the specialist feeding behaviour of the weevil Antliarhinus zamiae feeding on the seeds of Encephalartos spp. Furthermore, taller plants tend to bear more predated seeds, potentially because taller plants are easy located by the weevil. Finally, large canopy correlates negatively with predated seeds, mirroring our finding of increased seed predation in open habitats. Our SEM explains 67% of the variations in pre-dispersal seed predations, suggesting that this metamodel provides insights into the predisposition of cycad seed to predations. Since open habitats correlate with more seed predation, we suggest that anthropogenic activities that contribute to open forest must be avoided if we are to limit seed predations.