Estimation and analysis of insect population dynamics parameters via physiologically based models and hybrid genetic algorithm MCMC methods
Author(s)
Date Issued
2023
Type
article
Volume
77
Journal
Abstract
Decision support systems are gaining importance in several fields of agriculture, forest, and ecological systems management. Their predictive potential, entrusted to mathematical models, is of fundamental importance to set up opportune strategies to control pests and adversities that may occur and that may seriously compromise the natural equilibria. Among the others, population dynamics is one of the crucial challenges in the field. Despite the scientific community in recent years providing valuable models that faithfully represent terrestrial arthro- pods populations, such as insects, one of the main concerns is still represented by the parameter estimation. Parameters, in fact, characterise the species and their estimation are often entrusted to dedicated laboratory experiments that require specific equipment and highly qualified personnel. In this study we propose a novel method to estimate the model parameters directly from field data, where experimental activities are less expensive and less time consuming. In this study we propose a combination of least squares methods via genetic algorithms to preliminary evaluate the best parameter values and Markov Chain Monte Carlo approach to obtain their distribution. The algorithm has been tested in the special case of Drosophila suzukii, to quantify part of the parameters of an almost validated model in two steps: i) a first pseudo-validation using perturbed numerical solutions, and ii) a validation using real field data. The results highlighted the potentialities of the algorithm in estimating model parameters and opened several perspectives for further improvements from both the compu- tational and experimental point of view.
File(s)![Thumbnail Image]()
Name
1-s2.0-S1574954123002613-main.pdf
Description
Pubblicazione completa
Size
1.64 MB
Format
Adobe PDF
Checksum (MD5)
9588630664b1ba1ce66ed34ed738e66c