Amphibian communication networks and social organisation

Project leader: Max Ringler

Collaborators: Robert Höldrich (IEM/KUG), Palmyre Boucherie (Univ. Vienna), Eva Ringler (Univ. Bern)

Funded by the Austrian Science Fund (FWF):


In the evolution of social behaviour in vertebrates, amphibians take a paradoxical position, as to date no form of so-called ‘higher’ social organisation (e.g. living in social groups, persistent social structures, helper systems, coordinated foraging) is known in any amphibian. This is remarkable, because ‘higher sociality’ occurs in all other major vertebrate taxa. It is featured by several fish species and even more by many tetrapods – mainly birds and mammals, but to a lesser extent also by some reptiles. We hypothesize, that also in amphibians some degree of social organisation is present; that it should evolve sooner or later; or probably is currently in statu nascendi – as it is suggested by the prevalence of several ‘building blocks’ of social and communicative complexity in this group. Such adaptations include individual and kin recognition, structured communication, advanced orientation, temporal and/or spatial memory, prolonged territorial behaviour, high mate selectivity, elaborate mate choice mechanisms, and parental care.

In this project, we want to investigate the level of complexity of amphibian social organisation and thereby also establish a new, amphibian, model species for social behaviour research. We believe that investigating the structures and dynamics of frogs’ communication networks will provide valuable insights to better understand if, and how, frogs are socially organised. We start this investigation in the, in our opinion, most promising taxon – Neotropical poison frogs (Dendrobatidae) – with the intensively studied Brilliant-thighed Poison Frog Allobates femoralis. We employ state-of-the-art acoustic recording technology with sound source localization and individual recognition, combined with field observations, to record and quantify the full acoustic communication of an entire population of free ranging poison frogs. Using social network analysis, we will describe the communication network and evaluate its robustness through time and demographic changes. In a second step, using genetic sampling and pedigree estimation, we will measure the individual reproductive success and correlate it with individual calling activity and position in the CN to determine the role of calling in mate choice, and to identify related fitness effects and selective pressures.


Sound source localisation


Monitoring the calling activity of an entire poison frog population is only possible with passive acoustic recorders, utilizing sound source localisation, and individual recognition of individuals. To this is end, we are currently developing “SonicSpotter”, a large-scale wireless microphone array based on commercially available hardware (Raspberry Pi & ReSpeaker), in combination with a user-friendly software with a graphical user interface to conduct sound source localization and identify individual calling frogs. The development of the SonicSpotter is a collaboration with the Institute of Electronic Music and Acoustics at the University of Music and Performing Arts, Graz, Austria, and the two companies “sonible” (hardware development; and “atmoky” (software development; which are specialised in the reproduction and signal analysis of spatial audio. The system has currently undergone the first field-testing season and a second generation of SonicSpotter nodes has been developed based on field experiences. We are looking forward to deploy the first productive Spotters in our next field season in April-May 2023.


Figure 1: Territorial setup and reproductive behaviour of A. femoralis. Males call in stable, long term territories while females occupy perches between them. Females repeatedly visit males to mate and return to their perches. Males transport tadpoles to waterbodies and return to their territories.
Figure 2: Concept of a large scale wireless microphone array, with five ReSpeaker sensor nodes and direction of arrival estimation of two calling frogs. The heatmap represents a likelihood estimation of sound source location.
Figure 3: Field testing of the first generation of SonicSpotter recorders in spring 2022 in Camp Pararé, French Guiana.
Figure 4: Daniel Hojka (sonible) during synchronisation tests of the SonicSpotter recorders in spring 2022 in Camp Pararé, French Guiana.
Figure 5: Tests with the SonicSpotter mk2 recorders in fall 2022 in Graz, Austria.
Figure 6: Improved SonicSpotter mk2 recorder
Figure 7: Grafical user interface of the SonicSpotter software from atmoky with test recordings from Graz (sound source localisation and identification algorithms not yet implemented).




Prof. Robert Höldrich, IEM/KUG