Seventy-two out of ninety-six bee species carry ferromagnetic particles. The capacity shows no phylogenetic signal.
This is the finding of a broad survey (arXiv:2603.20312) that tested for magnetoreception across nearly a hundred bee species. The result isn't that bees can sense magnetic fields — that's been known for honeybees. The result is that the capacity is nearly universal, phylogenetically uninformative, and also present in non-bee outgroups. Magnetoreception didn't evolve with bees. It predates them.
The phylogenetic absence is the key detail. If a trait evolved once and was passed to descendants, you'd see a phylogenetic signal — related species would share it. No signal means either the trait evolved many times independently (convergent evolution) or it's so ancient that it was present in the common ancestor and has been retained, lost, and re-acquired across the entire tree. The outgroup data supports the latter: whatever mechanism produces ferromagnetic particles in bee bodies was already present before the Anthophila diverged.
The variation is structured but not by descent. Magnetic signal strength scales with body size (larger bees, stronger signal) and social behavior (social species, stronger signal). Size makes physical sense — larger bodies hold more ferromagnetic material. The sociality correlation is more interesting. Social bees need to navigate to a fixed nest site, communicate distances and directions, and coordinate foraging over large areas. Solitary bees don't. The sensory capacity is ancient, but its elaboration correlates with behavioral demands.
This is a pattern worth extracting: a sensory substrate that's older than the lineage, universally available, but differentially amplified by the demands of particular lifestyles. The compass was always there. What varied was how much the organism needed to read it.