Evidence of a sharp decline in insect biodiversity is accumulating. This is prompting us to step up our efforts to monitor changes in their biodiversity and abundance and to investigate the drivers of this decline. However, imprecise species identification, often due to a lack of human expertise, hampers our ability to accurately monitor insect decline. DNA-based identification methods have therefore been proposed as an effective and rapid way to conduct large-scale biodiversity monitoring. Yet, these methods cannot provide abundance estimates and are generally associated with systemic biases, which are rarely taken into account when assessing biodiversity.

To overcome these obstacles, we have developed a rapid, cost-effective and reliable framework (HAMI). It is based on a combination of bioinformatics and para-taxonomic expertise, achieved by performing a visual reconciliation of data between molecular results and high-resolution photographs of specimens. It has been first developed and tested on 500 samples of a highly diverse pilot group (Coleoptera), which came from a standardised biodiversity monitoring initiative covering more than 500 agricultural parcels in mainland France. It was then applied to no less than 1,000 additional samples from the same monitoring initiative to infer the impact of agricultural practices on the French Coleoptera community.

In two years of large-scale monitoring, by combining molecular and parataxonomist expertise, we were able to recover accurate species-level identifications for approximately 40,000 specimens from over 600 beetle species. HAMI also allowed us to reduce an average overestimation of 28% of the specific richness estimated by the molecular approach alone, due to sample cross-contamination and barcode database inaccuracies. Furthermore, HAMI allowed us to recover 22% of the missed specific composition of samples related to PCR bias. Otherwise, ecological analyses are still in progress, but preliminary results underline the importance of landscape heterogeneity in promoting beetle biodiversity.

Although molecular methods have the potential to perform large-scale insect monitoring, our results underline the fact that their use alone leads to a sizeable level of inaccuracy. By combining molecular approaches with parataxonomist expertise, HAMI significantly improved biodiversity estimates since reliable species identification can be obtained rapidly together with relative abundance data. Also, by optimizing the parataxonomist's time investment in key check points, HAMI enables to target gaps in barcode databases as well as problematic barcodes thus engaging in a virtuous circle of improvement of the curation of the molecular databases.