1. New insights on the behaviour of alternative types of individual-based tree models for natural forests.
- Author
-
Häbel, Henrike, Myllymäki, Mari, and Pommerening, Arne
- Subjects
- *
SCOTS pine , *DOUGLAS fir , *TREE populations , *GEOGRAPHIC spatial analysis , *REGRESSION analysis , *FOREST dynamics - Abstract
• Synoptic analysis of two IBM types yielded interesting details on similarities and differences. • New methodology of recombining model components was crucial to analysis. • Models with detailed interaction structure are only beneficial if the data provide the interaction information required. • Substantial progress in estimating the parameters of spatial IBMs was made. Agent/individual-based models (A/IBM) help to explain in a mechanistic way how spatial plant patterns evolve through time. In the past, seemingly different and independent types of A/IBMs were developed for modelling the dynamics of tree populations, e.g. growth interaction (GI) and shot noise (SN) models. In this paper, we present a new, advanced methodology of pattern-oriented modelling (POM) for the comparative, synoptic analysis of the behaviour of different types of A/IBMs by using recombinations of model components, validation and sensitivity analysis. We analysed model behaviour for spatio-temporal data from natural forests of interior Douglas fir (Pseudotsuga menziesii var glauca (Mirb.) Franco) and Scots pine (Pinus sylvestris L.) populations from Canada and the UK, respectively. Our detailed analysis clarified that both models, GI and SN along with their recombinations performed similarly and belong to the same group of A/IBMs. From the application of our new methodology we learnt that SN models were able to describe interactions more accurately than GI models and additionally produce interaction fields that can be used for other modelling purposes. On the other hand the GI model was more robust when using observed data that did not include sufficient information on tree interactions. Maximum-likelihood estimations were more reliable in spatial regression analysis than least-squares methods and should be preferred in spatial A/IBM parametrisation. [ABSTRACT FROM AUTHOR]
- Published
- 2019
- Full Text
- View/download PDF