Ecological Modelling 368 (2018) 377–390
Contents lists available at ScienceDirect
Ecological Modelling
journa l h om epa ge: www.elsevier.com/locate/ecolmodel
Global sensitivity analysis of a dynamic vegetation model: Model
sensitivity depends on successional time, climate and competitive
interactions
Nica Huber
*
, Harald Bugmann, Valentine Lafond
Forest Ecology, Institute of Terrestrial Ecosystems, Department of Environmental Systems Science, ETH Zurich, Universitätsstrasse 16, CH-8092 Zurich,
Switzerland
a r t i c l e i n f o
Article history:
Received 24 April 2017
Received in revised form
24 November 2017
Accepted 15 December 2017
Keywords:
Europe
ForClim
Forest gap model
Morris screening method
a b s t r a c t
Accurately predicting the dynamics of tree species productivity as well as their ranges at large scales is of
key importance for assessing the impact of global change on forests. Dynamic vegetation models (DVMs),
particularly forest gap models (FGMs), have been suggested as suitable tools for such joint predictions.
However, DVMs generally feature a large number of parameters that need to be estimated and may cause
considerable uncertainty in model outputs. In addition, model sensitivity may depend on environmental
conditions, stand composition and development stage. We systematically evaluated the parameter sen-
sitivity on simulated basal area of the state-of-the art FGM ForClim along a wide ecological gradient to
analyze model behavior and identify key parameters and processes that cause the highest variability in
model output. We applied the revised Morris screening method at 30 representative sites across Europe,
and compared results for monospecific and mixed stands at two system states in time, i.e. early and late
succession (dynamic equilibrium). The most influential parameters were related to tree establishment,
the water and light regimes, growth and temperature, whereby the relative parameter influence of the
latter strongly varied with climate. Further, model sensitivity differed between monospecific and mixed
stands as well as between early and late succession, reflecting the differential influence of ecological
processes with stand structure. We conclude that the parameter sensitivity of complex models should be
analyzed individually for several system states of interest. We recommend to focus the further develop-
ment (process representation and calibration) and analysis of FGMs on process representations related
to establishment, water limitations and phenology to improve the robustness of model predictions. We
provide recommendations for specific improvements of FGMs to better represent range dynamics.
© 2017 Elsevier B.V. All rights reserved.
1. Introduction
Accurately predicting tree species productivity as well as tree
species ranges and range shifts at large scales is of key impor-
tance for assessing the impact of global change on ecosystems
and the multiple ecosystem services they provide (e.g., Hanewinkel
et al., 2013; Montoya and Raffaelli, 2010). Climate change has been
shown to affect the geographical range of tree species (Walter
et al., 2002) and productivity (Boisvenue and Running, 2006). Since
forests are long-lived ecosystems, it is of key interest to understand
*
Corresponding author.
E-mail addresses: nica.huber@usys.ethz.ch (N. Huber),
harald.bugmann@usys.ethz.ch (H. Bugmann), valentine.lafond@usys.ethz.ch
(V. Lafond).
transient dynamics (i.e., non-equilibrium situations) that result
from changes in environmental drivers, succession, and large-
scale disturbances. Moreover, while studying species range shifts,
impacts of interspecific competition should be considered instead
of focusing on climatic changes only (e.g., Araujo and Luoto, 2007;
Brooker et al., 2007; Bullock et al., 2000; Meier et al., 2012; Nieto-
Lugilde et al., 2015).
Currently, two broad approaches are used to model for-
est productivity and plant distributions, namely correlative
and process-based approaches. To model plant distributions,
correlative species distribution models (SDM) relate species’
presence-absence data to environmental predictor variables (Elith
and Leathwick, 2009; Guisan and Zimmermann, 2000) and use
these correlations to project species’ occurrence probabilities
under changing conditions. They however assume that species dis-
https://doi.org/10.1016/j.ecolmodel.2017.12.013
0304-3800/© 2017 Elsevier B.V. All rights reserved.