Collective chemotactic localization emerges from interaction-driven phase transitions under temporally correlated noise

Chemotactic microorganisms operate in environments where signals are not only noisy but temporally structured, with finite correlation times that can severely impair gradient sensing and target localization. While previous models have extensively characterized the effects of fluctuating environments on individual chemotaxis, most theoretical frameworks treat agents as non-interacting, leaving unresolved how inter-bacterial interactions reshape collective robustness under temporally correlated noise. Here, we introduce a two-dimensional agent-based model of interacting run-and-tumble bacteria navigating noisy chemotactic landscapes. We show that short-range isotropic cohesion induces a two-stage collective response: interactions first stabilize population connectivity and above a finite interaction threshold, this structural cohesion translates into robust target localization even in regimes where individual chemotaxis fails. The resulting transition reveals an intermediate phase of cohesive but weakly localized states, demonstrating that structural condensation and functional targeting are distinct collective observables. We further demonstrate that selective heterotypic interactions in binary populations produce a structurally distinct collective regime characterized by dynamically maintained red-blue contact networks composed of transient mixed dimers and local heterotypic motifs. Unlike isotropic cohesion, selective interactions reorganize local contact topology without generating macroscopic condensation. These structures are quantitatively characterized through bond density, mixing statistics, and anisotropy metrics, and are governed primarily by interaction specificity rather than by environmental noise persistence. Together, these results establish that collective chemotactic behavior is controlled by the interplay between temporal signal correlations and interaction topology. More broadly, in this work we identify collective localization and internal organization as partially independent emergent properties of interacting active matter under fluctuating environments.