In Bahía de la Ascensión, Mexico, 'casitas' (large artificial shelters) are extensively used to harvest Caribbean spiny lobsters Panulirus argus. After the discovery of a pathogenic virus, Panulirus argus virus 1 (PaV1), in these lobsters, laboratory experiments revealed that PaV1 could be transmitted by contact and through water, and that lobsters avoided shelters harboring diseased conspecifics. To examine these issues in the context of casitas, which typically harbor multiple lobsters of all sizes, we examined the distribution and aggregation patterns of lobsters in the absence/presence of diseased conspecifics (i.e. visibly infected with PaV1) in 531 casitas distributed over 3 bay zones, 1 poorly vegetated ('Vigía Chico', average depth: 1.5 m) and 2 more extensively vegetated ('Punta Allen': 2.5 m; 'Los Cayos': 2.4 m). All zones had relatively high indices of predation risk. Using several statistical approaches, we found that distribution parameters of lobsters were generally not affected by the presence of diseased conspecifics in casitas. Interestingly, however, in the shallower and less vegetated zone (Vigía Chico), individual casitas harbored more lobsters and lobsters were actually more crowded in casitas containing diseased conspecifics, yet disease prevalence was the lowest in lobsters of all sizes. These results suggest that (1) investment in disease avoidance by lobsters is partially modulated by local habitat features, (2) contact transmission rates of PaV1 may be lower in nature than in the laboratory, and (3) water-borne transmission rates may be lower in shallow, poorly vegetated habitats more exposed to solar ultraviolet radiation, which can damage viral particles.