In collaboration with partners EBD-CSIC and the Technical University of Madrid (UPM), the ANTENNA project just launched the Biodiversity Forecasting Challenge - an interactive web application that puts human ecological intuition head-to-head with state-of-the-art machine learning. Users predict the future abundance of insect species from historical time series graphs, with access to contextual data on temperature, location, humidity, and species characteristics to inform their forecast.

After submitting via an intuitive slider, participants see how their prediction compares to the real recorded outcome and to other users worldwide. The challenge is not easy: insect populations show large year-to-year fluctuations that can confound even well-trained models. The platform’s modelling framework uses Reservoir Computing, developed at UPM, and all species data comes from the BIOTIME database, a public global repository of biodiversity time series. Species descriptions are AI-generated and reviewed by ecologists.

Each session ends with personalised performance feedback, and the challenge is replayable, featuring a different species every time. No prior expertise is needed. The application is free and accessible via the ANTENNA website here.

About ANTENNA: The goal of the research project is to fill key monitoring gaps through advancing innovative technologies that will underpin and complement EU-wide pollinator monitoring schemes, and to provide tested transnational pipelines from monitoring activities to curated datasets and enhanced indicators that support pollinator-relevant policy and end-users.