The integration of hierarchical levels of scale in tree species distribution models of silver fir (Abies alba Mill.) and European beech (Fagus sylvatica L.) in mountain forests

Alois Simon, Klaus Katzensteiner and Gudrun Wallentin

Ecological Modelling, 2023, vol. 485, issue C

Abstract: Taking into account and combining different levels of scale is essential in species distribution models when it comes to the investigation of the environmental niche of a species. The study focuses on two levels of spatial extent, on the one hand the continental level of Europe, and on the other hand the regional level of the federal state of Tyrol in Austria. Mountain forests in the European Alps cover the inner Alpine distribution margins of several tree species and therefore are particularly well-suited to reveal the predictive power of climate, soil, and topographic variables on the shaping of these margins. The potential occurrence of the two investigated tree species, Abies alba and Fagus sylvatica, is an important criterion for the delineation of elevational vegetation zones in forest site classification systems and was modelled with Deep Neural Networks. In the process, we observed a strong imbalance of absence and presence records at the continental level and evaluated different methods to address this issue. The potential predictor variables for species distribution modelling at the different spatial extents were grouped into climate, soil, and topographic information. The combination of the different hierarchical levels of extent and associated spatial resolution was implemented by using the outputs of the continental model as a predictor variable in the regional model. The binary classification of the 30% test dataset showed a True Skill Statistic of 0.73 to 0.76 for the regional level and 0.5 to 0.74 for the continental level, with slightly higher values for F. sylvatica than for A. alba. For both species and extent levels, the climate predictor group showed the greatest contribution (81 to 96%) to the models’ predictive power. At the regional level, climate was followed by soil and then topographic predictor groups. At the continental level however, topography showed stronger effects than soil information. In most cases, the consideration of soil information along climatic gradients led to an increase in the occurrence probability at the climatic distribution margins. There is evidence that soil conditions are more important in determining the inner Alpine distribution margins for F. sylvatica than for A. alba. To improve species distribution models at the regional level, e.g. in the Alpine area, a focus on soil information is proposed. In general, models which combine continental and regional data are preferable.

Keywords: Species distribution modelling; Cross-scale approach; Environmental predictor variables; Partitioning effects; Soil information (search for similar items in EconPapers)
Date: 2023
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0304380023002296
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:ecomod:v:485:y:2023:i:c:s0304380023002296

DOI: 10.1016/j.ecolmodel.2023.110499

Access Statistics for this article

Ecological Modelling is currently edited by Brian D. Fath

More articles in Ecological Modelling from Elsevier
Bibliographic data for series maintained by Catherine Liu ().