Modelling Plant Ecology - Environmental Modelling: Finding Simplicity in Complexity - page 202

vegetation models specify a direct empirical relationship
between soil moisture and the conductance of stomata to
water loss and CO 2 uptake (Jarvis, 1976; Cox et al ., 1998;
Sitch et al ., 2003; Woodward and Lomas, 2004; Krinner
et al ., 2005), however, the parameters of this empirical
approach are very poorly constrained. Theoretical exam-
inations of the maximum rate of water transport through
soils and plant water-conducting vessels (xylem) indicate
that, for a given combination of soil types, moisture con-
tent, and xylem vessel vulnerability to rupture under low
pressure, there is a maximum rate of water extraction,
beyond which, the conductance to water will fall to zero
at some point along the soil-leaf continuum (Sperry et al .,
1998). These analyses predict an 'envelope', representing
the maximum rate of water use which is physically possi-
ble. This approach is consistent with the often observed
minimum 'threshold' of leaf-water content which is main-
tained in many species (Tardieu and Simmoneau, 1998;
Fisher et al ., 2006; O'Grady et al ., 2007; McDowell et al .,
2008). Therefore, some models of plant water use and
stomatal conductance assume, again using an optimality
approach, that the conductance of stomata is regulated by
two competing processes; the need to keep leaves above
their minimum water content threshold, and the need to
maximize the rate of photosynthesis (Williams et al ., 1996,
2001; Buckley, 2005; Hickler et al ., 2006). These models
have met with some success in representing gas and energy
exchange in water-limited ecosystems (Williams et al .,
1996, 1998, 2001; Law et al ., 2000; Misson et al ., 2004;
Zeppel, 2008) but questions still remain over other types
of plant which demonstrate 'anisohydric' behaviour. In
these plants, regulation of leaf-water potential varies as
soil moisture changes (McDowell et al ., 2008) suggesting
that the 'optimal' approach may require adjustment for
other requirements of plant-water-use regulation. Dewar
(2002) suggested that the impact of plant-signalling com-
pounds on stomatal conductance may be required to
explain some patterns of water use that do not conform
to those predicted by the 'maximum water-use envelope'
approach to account for different strategies regarding the
acceptable safety margins of plant-water use.

expanded into providing information on the likely
outcome of changing climate and CO 2 concentrations on
changes in land-carbon storage and hence atmospheric
CO 2 . This development means that, in addition to
calculating gas and energy exchange, they must predict
the distribution and ecophysiology of different types of
vegetation. These requirements led to the development
of Global Dynamic Vegetation Models (Cox et al ., 2000,
Sitch et al ., 2003; Bonan et al ., 2003; Woodward and
Lomas, 2004; Krinner et al ., 2005), which attempt to
simulate the current distribution of the major 'biome'
types and how they may change in the future.
Several simplifying assumptions are used in the genera-
tion of DVMs. The most profound of these is the concept
of 'climate envelopes', which are based on the observation
that vegetation type is, at a large scale, defined by climate,
and that areas of the world with disparate geography
and plant phylogenies, but similar climate, converge on
similar 'functional types' of plant (Woodward, 1987).
Therefore, it is possible to describe the climate in which
given type of plant can survive, and from this, re-create
the global distribution of vegetation types. Most DVMs
in use in coupled carbon-climate models (Friedlingstein
et al ., 2006; Sitch et al ., 2008) use this methodology for
defining the range of establishment of different plant
functional types. While it is a useful simplification, this
highly empirical approach to understanding the controls
on plant community structure suffers when extrapolated
into situations not entirely analogous to the present
day - i.e.

when atmospheric CO 2 concentrations are

much higher, as predicted for the next century.
Another simplification used in DGVMs is Beer's
Law - a function that describes the typically exponential
decay of light levels through leaves that are scattered
evenly in space (Amthor, 1994). Again, the adaptive
ability of plants is represented here as the tendency for
leaf area to be distributed evenly in space, as plant-growth
signalling pathways act to place leaves in positions where
their light interception is maximized (if light is limiting).
Furthermore, first-order consideration of the optimal
distribution of nitrogen through a forest canopy dictates
that, if leaf photosynthetic capacity and light are to be
equally limiting, such that no resources are wasted, plants
should allocate photosynthetic capacity in proportion to
light availability. In this case, the Beer's Law function can
be integrated such that only one leaf layer is represented
in the simulation. This 'big leaf' simplification (Amthor
et al ., 1994) is traditionally used to scale between leaf-
and canopy-level fluxes in DGVMs but, as demonstrated
by Mercado et al . (2007), the profile of nitrogen through

Dynamic vegetation models
The use of SVAT models in climate models was initially
motivated by the need to provide information on the
division of energy into latent and sensible heat fluxes at
the surface, and the need to modify the surface albedo
for vegetation cover (Sellers et al ., 1986). However,
since then, the use of these models has been iteratively