Modelling Plant Ecology - Environmental Modelling: Finding Simplicity in Complexity - page 204

1996; Gross, 2008) is that tradeoffs between species traits
can account for observed patterns of biodiversity, as no
single species can use all resources optimally. Ecological
spaces that have differing resource availability are there-
fore likely to be occupied by different species (Tilman
and Pacala, 1993; Chesson and Huntley, 1997). Tradeoffs
may occur between the use of different limiting resources
(Tilman, 1982), between resource use and tolerance of
abiotic stress or predation (biotic stress), and between
the ability to colonize an area initially, or to compete
once established. The more tradeoff surfaces and the
higher the spatial or temporal heterogeneity in resource
availability, the more species can theoretically co-exist
(Gross, 2008). However, when actually measured, trade-
offs between different plant properties are rarely as exact
as mathematical models predict they need to be in order
to allow co-existence and most mathematical models do
not predict as many species as are observed in actual
ecosystems (Kneital and Chase, 2003). To reconcile the
conflicting theory and data, Clark et al . (2007) propose
that plant co-existence is a higher dimensional problem
than normally realized in models (Pacala et al ., 1996) and
that plants differ in many ways that we either do not rou-
tinely measure or do not understand. These unmeasured
axes of variation allow more species to co-exist than is
possible using a typical low-dimensional model. Further-
more, Clark et al . (2004) suggest that the variation within
species also acts to stabilize co-existence, compared to the
behaviour expected from a model which simply repre-
sents the mean properties of each species (Figure 12.3).
Therefore, in order to simulate the co-existence of differ-
ent species at either a stand- or landscape-level properly,
we must represent the parts of biological variation that
we do not understand.
From the point of view of modellers attempting to pre-
dict the response of plant species to climate change, these
issues represent a serious impediment. Theoretically, to
construct a model that may represent functional plant
diversity, it is necessary to measure enough traits from a
wide range of plants that an empirical representation of
all the relevant trade-off axes is achieved, but such efforts
are currently some way from achieving this goal (Lavorel
et al ., 2007). A potentially more accurate but relatively
embryonic method of generating the range of possible
plant species is to determine the interrelations between
plant attributes based on the optimizing principle of
competitive evolution (Westoby, 2006). This approach,
along with further research into the costs and benefits of
different types of plant construction (Hacke et al ., 2006),
holds great potential for redefining the way in which

plant properties are theoretically represented, although,
currently, it is only in the embryonic stage. Early efforts
to adopt this logical framework into vegetation models
are represented by Franklin (2007). In this model of the
responses of vegetation to CO 2 enrichment, canopy nitro-
gen content and leaf area index are optimized in order
to maximize the carbon available for vertical growth rate
and reproduction, the properties considered analogous to
evolutionary 'fitness'. This optimization is based on the
principle that nitrogen has benefits of higher photosyn-
thetic rates, and costs of maintenance respiration required
to keep the nitrogenous proteins active. Photosynthesis is
nonlinearly related to total canopy nitrogen because of the
effects of self-shading, while respiration is linearly related.
This relationship means that at some leaf area there is a
compensation point where leaves cost more to construct
and maintain than they produce in photosynthate. The
Franklin model finds a closed form equation to predict
the 'optimal' properties that maximize plant fitness. In
this manner, some plant traits may be predicted from the
'bottom' up, rather than empirically measured; a poten-
tially more elegant approach to extrapolating the results
of CO 2 fertilization experiments into future climates.

12.4 Case study

The impact of climate change on the Amazon rainforest
carbon balance to climate change has been identified as a
critical 'tipping point' in the Earth system (Lenton et al .,
2008) on account of the predictions that temperatures
will increase by 1.8 to 5.1 ◦ C, dry season rainfall will
decline over the next century, which may lead to sub-
stantial changes in vegetation cover (Malhi et al ., 2008).
Even small changes in vegetation cover and evaporation
rates are likely to be amplified by increasing fire risk
and decreases in regional rainfall (30-50% of which is
recycled within the Amazon), further exacerbating the cli-
mate changes (Eltahir and Bras, 1994; Oyama and Nobre,
2003). Most predictions of the impact of changing cli-
mate on rainforest function are based on models of the
response of ecosystems to soil drought which have never
been tested against empirical data, largely because existing
measurements of land-atmosphere exchange in Amazo-
nian rainforests appear to have only provided one example
of a forest briefly experiencing water limitation thus far
(Malhi et al ., 1998; Carswell et al ., 2002; Saleska et al .,
2003; Goulden et al ., 2004; Harris et al ., 2004; Loescher
et al ., 2005). Furthermore, most widely used models pre-
dict large declines in forest evaporation caused by drought