Extrasolar planets with sizes between that of the Earth and Neptune ($R_{rm p}=1{-}4~{rm R}_oplus$) have a bimodal radius distribution. This planet radius valley separates compact, rocky super-Earths ($R_{rm p}=1.0{-}1.8~{rm R}_oplus$) from larger sub-Neptunes ($R_{rm p}=1.8{-}3.5~{rm R}_oplus$) hosting a gaseous hydrogen-helium envelope around their rocky core. Various hypotheses for this radius valley have been put forward, which all rely on physics internal to the planetary system: photoevaporation by the host star, long-term mass loss driven by the cooling planetary core, or the transition between two fundamentally different planet formation modes as gas is lost from the protoplanetary disc. Here we report the discovery that the planet radius distribution exhibits a strong dependence on ambient stellar clustering, characterised by measuring the position-velocity phase space density with textit{Gaia}. When dividing the planet sample into field and overdensity sub-samples, we find that planetary systems in the field exhibit a statistically significant ($p=5.5times10^{-3}$) dearth of planets below the radius valley compared to systems in phase space overdensities. This implies that the large-scale stellar environment of a planetary system is a key factor setting the planet radius distribution. We discuss how models for the radius valley might be revised following our findings and conclude that a multi-scale, multi-physics scenario is needed, connecting planet formation and evolution, star and stellar cluster formation, and galaxy evolution.
Download