Wild bees are essential pollinators and therefore play a key role in both natural and agricultural ecosystems. However, bees have often been neglected in conservation studies and policies worldwide, which is surprising given their ecological importance. As a result, little is known on the conservation status of the vast majority of wild bee species in Europe, and even less worldwide. Limited surveys suggest important declines in the abundance and diversity of most wild bee communities worldwide. It is therefore urgent to implement targeted measures for the conservation of these keystone species. Once implemented, the effectiveness of these measures must be evaluated using adequate monitoring programs. To date, wild bee surveys are entirely based on morphological identification, which is both labor intensive and time consuming. Consequently, an affordable, high-throughput identification method is needed to reduce costs and improve bee monitoring. The objective of this thesis was to evaluate novel genetic techniques based on Next Generation Sequencing (NGS) methods for facilitating surveys of wild bees. NGS tools were mainly investigated for bridging two important impediments to wild bee conservation efforts, i.e., the cost of biodiversity assessment schemes and taxonomic incompleteness. With the development of NGS techniques, DNA barcoding has gained enormous momentum, enabling cost-effective, fast and accurate identifications. Before these methods can be routinely used in monitoring programs, there are however still important knowledge gaps to fill. These gaps mainly concern the detection of rare species and the acquisition of accurate quantitative data on species abundance; more generally the cost and labour effectiveness of these methods need to be evaluated. To provide a comprehensive presentation of the advantages and weaknesses of different NGS-based identification methods, we assessed three of the most promising ones, namely metabarcoding, mitogenomics and NGS barcoding. Using a regular monitoring data, we found that NGS barcoding performed best for both species’ presence/absence and abundance data, producing only few false positives and no false negatives. The other methods investigated were less reliable in term of species detection and inference of abundance data, and partly led to erroneous ecological conclusions. In terms of workload and cost, we showed that NGS techniques were more expensive than morphological identification with our dataset, although these techniques would become slightly more economical in large-scale monitoring programs. A second aim of this thesis was to provide an easy and robust genomic solution to alleviate taxonomical incompleteness, one of the major impediments to the effective conservation of many insect taxa. For conservation purposes, having stable and well-delimited species hypotheses is essential. Currently, most species are delimitated based on morphology and/or DNA barcoding. These methods are however associated with important limitations, and it is widely accepted that species delimitation should rely on multi-locus genomic markers. To overcome these limitations, ultraconserved elements (UCEs) were tested as a fast and robust approach using different species-complexes harbouring cryptic diversity, mitochondrial introgression, or mitochondrial paraphyly. Phylogenetic analyses of UCEs were highly conclusive and yielded meaningful species delimitation hypotheses in all cases. These results provide strong evidence for the potential of UCEs as a fast method for delimiting species even in cases of recently diverged lineages. Advantages and limitations of UCEs for shallow phylogenetic studies are further discussed.
Radiomics side experiments and DAFIT approach in identifying pulmonary hypertension using Cardiac MRI derived radiomics based machine learning models
