Background: Biology's dark matter is found in hyperdiverse, poorly known taxa that are numerically dominant and yet largely unstudied, even in temperate regions of the globe where charismatic taxa are already well understood. Biology's dark matter is everywhere, but identification impediments caused by high diversity, abundance, and small size have historically stymied its study. We here show how entomological dark matter can be converted from dark to reasonably well known by using high-throughput DNA barcoding ("megabarcoding"). We reveal the surprisingly high abundance and diversity of scuttle flies (Diptera: Phoridae) across Sweden by processing 31 800 specimens from 37 sites across four seasonal time-periods. We ask how many scuttle fly species are likely to occur in Sweden and what environmental factors are driving community changes between seasons and across the country. Results: We find that Swedish scuttle fly diversity is much higher than previously known, with 549 mOTUs (putative species) detected, compared to only 374 species recorded previously. Hierarchical Modelling of Species Communities reveals that scuttle fly communities are highly structured by latitude, and strongly driven by climatic (including temporal) factors. The large dissimilarities between sites and seasons are largely driven by turnover rather than nestedness (i.e. local differences in species richness). We predict that climate changes are likely to greatly affect the 29% of the species that show a positive and the 18% of species that show a negative response to mean annual temperature. Surprisingly, these results were robust regardless of whether we used haplotype diversity or species-proxies (mOTUs: molecular Operational Taxonomic Units) as response variables. Furthermore, species-level models of common taxa adequately predict overall species richness. Conclusions: Understanding the bulk of the diversity around us is imperative during an era of biodiversity loss and we here show that dark insect taxa can be efficiently characterised and surveyed with megabarcoding. Furthermore, we show that undersampling of rare taxa and the choice of operational taxonomic units do not change the main ecological inferences, so that the time appears rife for tackling biology's dark matter.