Biodiversity loss threatens ecosystems and human well-being, making accurate, large-scale monitoring crucial. Environmental DNA (eDNA) has enabled species detection from substrates such as water, without the need for direct observation. Lately, airborne eDNA has been showing promise for tracking organisms from insects to mammals in terrestrial ecosystems. Conventional biodiversity assessments are often labor-intensive and limited in scope, leaving gaps in our understanding of ecosystem response to environmental change. Here, we demonstrate that airborne eDNA can detect organisms across the tree of life, quantify changes in abundance congruent with traditional monitoring, and reveal land-use induced regional decline of diversity in a northern boreal ecosystem over more than three decades. By analyzing 34 years of archived aerosol filters, we reconstruct weekly temporal relative abundance data for more than 2700 genera using non-targeted methods. This study provides unified, ecosystem-scale biodiversity surveillance spanning multiple decades, with data collected at weekly intervals on both the individual species and community level. Previously, large scale analyses of ecosystem changes, targeting all types of organisms, has been prohibitively expensive and difficult to attempt. Here, we present a way of holistically doing this type of analysis in a single framework.