Top: photo of Habronattus agricola by Damian Elias.

Bottom: Confusion matrix showing tests of automated classification of signals from Habronattus agricola

Jumping spider communication

An enormous number of species communicate and navigate the world using substrate-borne vibration, including thousands of invertebrates. Unlike airborne communication, this realm is very understudied and researchers are only now beginning to understand the intricacies of how animals communicate through surface waves that move through substrates like leaves, branches, or sediment.

Jumping spiders don’t hear sound waves that move through they air, instead they sense all sound via vibrations that travel through surfaces. When courting potential mates, male spiders put on a song and dance routine that involves a series of specific substrate-borne sounds created by thumping their legs and abdomen on the ground (great overview here).

Working with the Elias Lab at UC Berkeley, I’ve been apply machine learning techniques to automatically classify the distinct sounds each spider makes. The goal is to be able to automatically characterize different species repertoires, which can help shed light on which signals are gained and lost as speciation occurs and species adapt to different environments.