# A web-based tool for rapid and accurate craniometric differentiation of clouded leopard species

## TL;DR

A free web tool was developed to accurately identify clouded leopard species using skull measurements, aiding wildlife crime investigations and museum curation.

## Contribution

A replicable statistical framework and web-based classifier for species differentiation using craniometric data.

## Key findings

- A categorical model using two skull metrics achieved 97% accuracy in species differentiation.
- A continuous predictor model using postorbital width achieved 80.6% accuracy for males and 85.6% for females.
- The web-based tool is free, user-friendly, and supports anti-trafficking and museum curation efforts.

## Abstract

The illegal wildlife markets of Southeast Asia are bolstered by organised criminal networks and the region’s rich density of charismatic wildlife. Forensic tools identifying species and their origins are vital to combat wildlife crime. However, many require expensive technology and skilled personnel, limiting their use in rural trade hotspots. This study introduces a replicable statistical framework, using skull morphometrics, to distinguish related species with simple measurements. We developed a web-based classifier trained on clouded leopard (Neofelis spp.) skulls from museum collections across Europe, Asia and the U.S.A., a genus often targeted in wildlife trade. Our categorical predictive model, based on two key metrics, the fronto-nasal “pit” and m1 talonid morphology achieved 97% accuracy (p < 0.005). A continuous predictor model, using postorbital width, achieved 80.6% accuracy for males and 85.6% for females (both p < 0.05). These models were encoded into a free, user-friendly web app, enabling practitioners in remote areas to distinguish these two species easily. This tool not only supports anti-trafficking efforts but also enables museum curators to correctly assign provenance to clouded leopard skulls with uncertain origins.

The online version contains supplementary material available at 10.1038/s41598-025-96080-w.