Quantifying avian pest control: developing the Functional Insectivory Index through a modular multisensory network for monitoring of bird predation

Descripción del proyecto

De las instantáneas al seguimiento ecológico continuo

Comprender cómo interactúa la fauna silvestre con los ecosistemas sigue siendo un reto importante, ya que los métodos tradicionales de seguimiento suelen ofrecer únicamente observaciones puntuales del comportamiento animal. En el proyecto ORNISCOPE, respaldado por las acciones Marie Skłodowska-Curie, se emplearán cámaras inteligentes con inteligencia artificial para llevar a cabo un seguimiento continuo de la actividad de aves en entornos naturales. Se desarrollará una red de cámaras de bajo consumo energético y un modelo de inteligencia artificial adaptado para detectar e identificar especies de aves. Este sistema permitirá automatizar el seguimiento de la fauna y estudiar cómo influye la estructura de los bosques en el control de plagas ejercido por las aves. Además, se pondrán a disposición hardware, software y conjuntos de datos de acceso abierto para facilitar el uso de técnicas avanzadas de seguimiento de la biodiversidad y reducir sus costes. En conjunto, los resultados de ORNISCOPE podrían contribuir a mejorar la conservación de la biodiversidad y la gestión sostenible del paisaje.

Objetivo

What if we could watch nature as it happens, not just in snapshots? ORNISCOPE turns snapshots into flow with energy-efficient smart cameras and server-side AI that run for days in the field at 2 frames per second.
The project bridges field ecology and smart sensing in a natural feedback loop:
(1) Model. In year one, we retrain a deep-learning model on a remote server using a large, curated image set so it precisely filters empty frames, flags bird activity, and identifies species.
(2) Network. Following the model retraining, we build a modular network of low-power cameras that capture and buffer images locally, then automatically send them to the server, where the customised model performs filtering and identification.
(3) Ecology. With this tool in place, during the second year, we conduct a field study across forests that differ in structure and management, pairing camera outputs with bird counts, insect sampling, and leaf-damage surveys.
From these data, we compute the Functional Insectivory Index (FII): a simple, behaviour-based indicator of natural pest control that merges bird abundance, trait-based potential to eat herbivores, and observed predation captured by cameras. FII enables researchers and managers to compare sites consistently and track how management affects biocontrol.
The project delivers new evidence on how forest structure and management history shape predator–prey dynamics and bird-mediated pest suppression, plus a reusable open-access toolset (hardware designs, software, datasets) that lowers the cost of precise, high-volume ecological monitoring. As an MSCA Postdoctoral Fellowship, it also provides advanced training in AI, field ecology, and open science, strengthening European capacity for nature management. By turning snapshots into flow, ORNISCOPE supports faster, better-informed decisions for a healthier, more resilient nature and smarter landscape management.