Early and Noninvasive Bird Gender Identification by ATR-FTIR Spectra Coupled with a Randon Forest Algorithm

TL;DR

This paper introduces a noninvasive and cost-effective method using ATR-FTIR and machine learning to identify bird gender early in life.

Contribution

The study proposes a novel, noninvasive method for early bird gender identification using ATR-FTIR and Random Forest, avoiding DNA testing.

Findings

ATR-FTIR combined with Random Forest achieved 94.4% accuracy for Bicudo and Curio birds.

The method showed 77.8% accuracy for Cockatiel and 72.2% for Ring-necked Parakeet.

PCA and SNV preprocessing improved the performance of the ATR-FTIR data analysis.

Abstract

The early identification of sex in birds is essential
for reproduction,
breeding programs, and commercialization and plays a crucial role
in wildlife management and environmental law enforcement. The DNA-based
molecular techniques, known for their accuracy and noninvasive nature,
are the primary methods for sex determination. However, these techniques
are time-consuming and expensive and require specialized laboratories.
This study explores the use of attenuated total reflectance Fourier
transform infrared spectroscopy (ATR-FTIR) combined with the Random
Forest algorithm as a noninvasive, cost-effective, and precise alternative
for early gender determination in birds. Measurements were performed
on the feather region, known as the vexillum, of four bird species: Oryzoborus maximiliani (Bicudo), Nymphicus
hollandicus (Cockatiel), Oryzoborus
angolensis (Curio), and Psittacula
krameri (Ring-necked Parakeet). The bird’s
sex was confirmed by DNA analysis. The ATR-FTIR spectra in the range
of 3800–800 cm–1 were processed using standard
normal variate (SNV) and analyzed with principal component analysis
(PCA) to reduce dimensionality and highlight significant transitions.
Processing ATR-FTIR spectra by the Random Forest classifier yielded
promising results, with accuracy rates in an external validation of
94.4% for Bicudo and Curio, 77.8% for Cockatiel, and 72.2% for the
Ring-necked Parakeet. These findings highlight the potential of ATR-FTIR
as a viable technique for the early identification of gender in birds.

Figures