High-throughput 3D phenotyping in northern quahogs Mercenaria mercenaria for dimensional trait measurement and weight prediction

Abstract

The northern quahog Mercenaria mercenaria (also called hard clams) is one major bivalve species for aquaculture in the US. Accurate shell geometry quantification is essential for genetic breeding. Traditional manual measurements of dimensional traits and live weight of hard clams are time-consuming and prone to errors, particularly for volume estimation. This study introduces a low-cost, high-throughput method for measuring hard clam dimensions and volumes using digital 3D replicas and predicting fresh weight through machine learning. The extracted shell length, height, and width from 3D point cloud models demonstrated high accuracy (R² ≥ 0.93). Although the extracted volumes showed some deviation from ground truth values (R² = 0.60), the method provided a higher resolution of volume and was significantly faster than manual measurements. By combining the three extracted dimensions and volumes, fresh weight was accurately predicted using linear regression (R² = 0.96). These results highlight the potential of the proposed method to significantly reduce labor costs while providing precise trait values to support the selective and genomic breeding of northern quahogs. Moreover, this method can be extended to accurate and fast phenotyping of other bivalve species, representing a substantial advancement in broodstock breeding technologies.

Keywords

• Mercenaria mercenaria
• Hard clams
• High-throughput phenotyping
• 3D replicas
• Shell metric traits
• Machine learning