Jump to Main Content
INSTAR: An Agent-Based Model that integrates existing knowledge to simulate the population dynamics of a forest pest
- Suárez-Muñoz, María, Bonet-García, Francisco, Hódar, José A., Herrero, Javier, Tanase, Mihai, Torres-Muros, Lucía
- Ecological modelling 2019 pp. 108764
- Thaumetopoea pityocampa, climate, forest pests, instars, land use, models, phenology, plantations, population dynamics, spatial variation, uncertainty analysis, Mediterranean region
- Pine plantations, very common in the Mediterranean basin, are recurrently affected by forest pests due to intrinsic characteristics (high density, low spatial heterogeneity) and external factors (consistent trend towards a warmer and drier climate). INSTAR is an Agent-Based Model aiming to simulate the population dynamics of the Thaumetopoea pityocampa forest pest. The model has been designed using a modular approach: several interconnected modules (submodels) facilitate the incorporation of new knowledge about the pest biology and can serve as template for the design of other similar models. The model is spatially and temporally explicit and allows its implementation under different climate and land use scenarios. INSTAR is described in detail in this manuscript using the standardized ODD (Overview, Design concepts, and details) protocol.Temperature is known to be one of the main factors modulating the population dynamics of T. pityocampa. In order to be coherent and structurally realistic, INSTAR should faithfully reproduce the effect of this factor on the species’ phenology. This requirement has been assessed here through a consistency test of the submodules responsible of species development. This assessment is constituted by a calibration analysis of the pest phenology and a stress test performed by exposing the model to extreme climate inputs. As a result of calibration, the model successfully reproduces the phenology of the species in the simulated study area. Moreover, the stress test confirmed that the model behaves as expected when exposed to extreme input values. The results presented in this manuscript constitute a first internal validation of the development submodels. After this, INSTAR is ready for a deeper analysis consisting on a sensitivity and uncertainty analysis.