Constraining environments constitute a strong selective pressure potentially driving strategies optimizing the reliability of communication processes. In this perspective, the “Signal Structure Hypothesis” predicts that the structure of animal signals will differ depending on features of the habitat. For example, in the acoustic domain, the so-called “Acoustic Adaptation Hypothesis” predicts that bird songs optimized for long-range propagation in a forest should be slowly modulated in frequency and as low in frequency as the sender can produce. Since the susceptibility of emitted signals to propagation-induced modifications depends on their acoustic characteristics, the emitter also could enhance or on the contrary reduce the active space of its communication by coding the information in more or less propagation-resistant parameters. I will show that this adjustment of coding strategies according to the habitat can be experimentally demonstrated for the vocalizations of songbirds living in different habitats. Due to the variable spacing of territorial individuals, some information may be coded to degrade over short distance and some might be coded such that it transmits over long distance without much degradation. Thus, species-specific identity is encoded in propagation resistant acoustic features allowing individual to reach a wide audience and thus constitutes public information. Conversely, group identity, individual identity and motivation state are encoded by song features susceptible to propagation, this private information being reserved for close neighbors or for the partner.