Behavioral Response of Porcellio Scaber Sensory Integration and Habitat Prediction Using Machine Learning

Soil macroinvertebrates play a crucial role in ecosystem functioning and are widely recognized for their sensitivity to environmental conditions, mak-ing them valuable bioindicators of soil quality. Among them, the terrestrial isopod Porcellio scaber is particularly relevant due to its close interaction with soil substrates and responsiveness to physical and chemical changes. This paper evaluates the influence of temperature, substrate type, and sali-nization on the habitat selection of P. scaber to assess its potential as a bi-oindicator in sustainable agricultural systems. Experiments were conducted in Petri dishes containing black soil combined with the different substrates under controlled temperature conditions. Habitat selection was defined as a binary classification outcome and used as the response variable in super-vised machine learning analyses. The dataset was analyzed using Decision Tree (DT) and Support Vector Machine (SVM) classifiers. The DT identified substrate configuration—particularly the presence of banana—as the prima-ry determinant of distribution patterns. Temperature was found to further structure substrate differentiation, while salinization was found to exert a secondary contextual effect. The SVM confirmed the combined predictive relevance of temperature, salinity, and substrate variables, indicating that habitat selection emerged from the interaction between trophic availability and abiotic conditions rather than from a single isolated factor. These findings show that P. scaber displays context-dependent habitat selection influenced by resource availability and environmental conditions, supporting its use as an ecological indicator in sustainable soil management.