The evolution of phenotype, computation, and network mechanism of song recognition in crickets

Jan Clemens - Georg-August-Universität Göttingen
Matthias Hennig - Humboldt-Universität Berlin

The diversity of animal behaviors – even among closely related species – is testament to the high evolutionary plasticity of the neural circuits driving them. How neural circuits evolve to produce diverse behavioral phenotypes is largely unknown. This is because the nonlinear mapping from circuit parameters to computation and behavior crucially shapes behavioral evolution yet remains unidentified. Acoustic communication systems exhibit highly diverse signal and recognition phenotypes and can diverge rapidly, suggesting highly evolvable and shared core circuits. Comparing recognition phenotypes, computational principles and circuit mechanisms in a phylogenetic context is therefore ideally suited to identify the evolutionary principles of circuit design.We propose to implement this approach in crickets. Males produce a species-specific song – pulse trains that are often grouped into chirps – analyzed by females before approaching the male. Male song and female behavior are highly quantifiable in a tractable parameter space, allowing a comprehensive description of the phenotype. Prior and recent work has identified the major computational principles for more than 20 species and a shared core circuit underlying the short-timescale phenotype. This sets the stage for assessing the phenotype, computation and mechanism of song pattern recognition in an evolutionary context. In a first step, we will systematically compare the computations producing the phenotypic diversity, by fitting existing computational models to the behavioral data of all species. Novel computations will be identified by focusing on species with unusual resonant phenotypes or complex songs. The computational principles of song recognition on the long timescale of chirps will be identified using the non-trivial interactions between timescales as a powerful constraint.Concurrently, we will use a model of the shared core circuit to determine if it can produce all the computations that generate the phenotypic diversity. This will identify key parameters driving circuit divergence and specialization. Model predictions will be tested in behavioral experiments and using electrophysiology across species. To extract evolutionary trends and trajectories we will assess the computations and mechanisms underlying the phenotypic diversity in a phylogenetic context. This will provide crucial insights into the circuit’s evolutionary plasticity based on the circuit’s parameter space. Moreover, the match between song and recognition template will identify the song features underlying stabilizing and directional selection, and the circuit parameters that govern their recognition. Overall, we will combine computational modeling, behavioral tests and electrophysiology to reveal the functional specialization of the computations and mechanism producing a homologous yet diverse behavior and provide important insights into the principles shaping neural circuit evolution.