Get access

On the identification of the most suitable traits for plant functional trait analyses

Authors

  • Markus Bernhardt-Römermann,

  • Christine Römermann,

  • Robert Nuske,

  • Andreas Parth,

  • Stefan Klotz,

  • Wolfgang Schmidt,

  • Jutta Stadler


M. Bernhardt-Römermann (markus.bernhardt@forst.uni-goettingen.de), A. Parth and W. Schmidt, Dept Silviculture and Forest Ecology of the Temperate Zones, Georg–August Univ. Göttingen, Faculty of Forest Sciences and Forest Ecology, Büsgenweg 1, DE–37077 Göttingen, Germany. – C. Römermann, Inst. of Physical Geography, Univ. of Frankfurt, Altenhöferallee 1, DE–60438 Frankfurt am Main, Germany. – R. Nuske, Dept Ecoinformatics, Biometrics and Forest Growth, Georg-August Univ. Göttingen, Faculty of Forest Sciences and Forest Ecology, Büsgenweg 4, DE–37077 Göttingen, Germany. – S. Klotz and J. Stadler, Dept of Community Ecology Helmholtz, Centre for Environmental Research – UFZ, Theodor-Lieser-Str. 4, DE–06120 Halle/Saale, Germany.

Abstract

Within the past few years plant functional trait analyses have been widely applied to learn more about the processes and patterns of ecosystem development in response to environmental changes. These approaches are based on the assumption that plants with similar ecologically relevant trait attributes respond to environmental changes in comparable ways. Several methods have been described on how to analyse a priori defined trait sets with respect to environment. Irrespective of the statistical methods used to contrast ecosystem responses and environmental conditions, each functional trait approach depends strongly on the initial trait set. In nearly all recent studies on functional trait analysis a test, if a trait is responsible, is applied independently from the core analysis. In the current study we present a method that extracts those traits from a wider set of traits which are optimal for describing the ecosystem response to a given environmental gradient. This was done by the use of iterative three-table ordination techniques with each possible trait combination. We further concentrated on the effect of the inclusion of too many traits in such analyses.

As examples the method was applied to three long term studies on abandoned arable fields. The approach was validated by comparing the results with literature-knowledge on arable field succession. Although the trait pre-selection was only based on a statistical procedure, our method was able to identify all relevant processes of ecosystem responses. All three sites show comparable ecosystem responses; the importance of the competitive ability of plants was highlighted. We further demonstrated that the use of too many traits results in an over-fitting of the trait-environment model.

The presented method of iterative RLQ-analyses is adequate to identify responding traits to environmental changes: the discovered processes of successional development of abandoned arable fields are consistent with our knowledge from the literature.

Get access to the full text of this article

Ancillary