New Root growth models: towards a new generation of continuous...
Transcript of New Root growth models: towards a new generation of continuous...
Journal of Experimental Botany, Vol. 61, No. 8, pp. 2131–2143, 2010doi:10.1093/jxb/erp389 Advance Access publication 27 January, 2010
REVIEW PAPER
Root growth models: towards a new generation ofcontinuous approaches
Lionel Dupuy*, Peter J. Gregory and A. Glyn Bengough
Scottish Crop Research Institute, Invergowrie, Dundee DD2 5DA, UK
* To whom correspondence should be addressed. E-mail: [email protected]
Received 24 August 2009; Revised 24 November 2009; Accepted 8 December 2009
Abstract
Models of root system growth emerged in the early 1970s, and were based on mathematical representations of root
length distribution in soil. The last decade has seen the development of more complex architectural models and the
use of computer-intensive approaches to study developmental and environmental processes in greater detail. There
is a pressing need for predictive technologies that can integrate root system knowledge, scaling from molecular to
ensembles of plants. This paper makes the case for more widespread use of simpler models of root systems based
on continuous descriptions of their structure. A new theoretical framework is presented that describes the dynamics
of root density distributions as a function of individual root developmental parameters such as rates of lateral root
initiation, elongation, mortality, and gravitropsm. The simulations resulting from such equations can be performedmost efficiently in discretized domains that deform as a result of growth, and that can be used to model the growth
of many interacting root systems. The modelling principles described help to bridge the gap between continuum and
architectural approaches, and enhance our understanding of the spatial development of root systems. Our
simulations suggest that root systems develop in travelling wave patterns of meristems, revealing order in otherwise
spatially complex and heterogeneous systems. Such knowledge should assist physiologists and geneticists to
appreciate how meristem dynamics contribute to the pattern of growth and functioning of root systems in the field.
Key words: Architecture, deformable model, development, dynamics, meristem, root, wave.
Introduction
Root system architecture is important for the plant to
access soil resources. There is now considerable evidence
linking root architecture with water and nutrient acquisition
efficiency (Hodge et al., 1999; White et al., 2005; Lynch,
2007). The structure and dynamics of root system architec-
ture, however, are complex, and architectural modelling has
resulted from the need to incorporate space into models andto take account of the biophysical interactions taking place
between roots and their environment. Simple spatial models
of root growth and development have been available for
a long time (Gerwitz and Page, 1974), but the advent of
more powerful computers and the availability of software
and simple programming languages has facilitated the
development of more complex plant architectural models
(Jourdan and Rey, 1997)—some examples of model typesare given in Table 1.
Currently, most plant root architectural models use
computer simulations to reproduce the developmental pro-
cesses of root apical meristems and to construct virtual root
architectures: single roots are assembled incrementally
through the growth of a set of virtual apical meristems
whose activity is determined at each time step of the
simulation. The resulting root architecture is a complexdata structure describing the geometrical properties of the
different roots and the topology of connection they
establish with each other. Usually the same morphogenetic
rules are used to prescribe the behaviour of whole sets of
meristems, and complex architectures arise as emergent
properties of these simple rules (Prusinkiewicz, 2004).
The last 10 years have seen the development of a variety of
models integrating important biological and physical pro-cesses of growth, a discipline now referred to as ‘functional
ª The Author [2010]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved.For Permissions, please e-mail: [email protected]
at Niedersaechsische S
taats- u. Universitaetsbibliothek G
oettingen on July 22, 2010 http://jxb.oxfordjournals.org
Dow
nloaded from
structural plant modelling’ (Godin and Sinoquet, 2005).
Root architectural models have been used to describe water
and nutrient transfer from soil to plants (Doussan et al.,
1998; Dunbabin et al., 2002). Elaborate models of light
interception by shoots and canopies have been developed
(Birch et al., 2003), and whole plant architectural models
that simulate carbon and mineral elements are beingconstructed (Drouet and Pages, 2007). Architectural models
have also proven to be particularly powerful in analysing
plant–environment interactions, incorporating various
forms of mechanical (Fourcaud et al., 2003; Dupuy et al.,
2005a) and hydraulic (Somma et al., 1998) processes.
Models incorporating links to genetic data (Tardieu, 2003;
Letort et al., 2008) and processes regulating development
are now under active development (Lucas et al., 2008).The success of models incorporating root system archi-
tecture has also created a need for different types of
models and approaches (de Dorlodot et al., 2007), and also
for realistic parameter values for a range of crops (Tsegaye
et al., 1995). For example, the models needed to assist
breeding strategies (Hammer et al., 2006) or population
dynamics involve simulating large numbers of single plants
simultaneously (Podlich and Cooper, 1998; Grimm, 1999).
This is still a huge computational challenge for architec-
tural models because the functioning of all organs is
computed explicitly (e.g. Dunbabin, 2007). Effective mod-
elling of plant–soil interactions also requires coupling
discrete structures, i.e. roots, to continuous descriptionsof the environments (e.g. soil mechanics and hydraulic
models). Other forms of spatial models that could provide
efficient solutions to these problems, continuous-based
approaches in particular, have not developed at a compa-
rable rate.
The purpose of this paper is to review existing, continu-
ous, spatial approaches for the modelling of root systems,
set them in a historical context, and to examine theirpotential to provide new insight into the mechanisms of
plant growth and development. Recent advances in the field
are presented and the case will be made that further
development of these techniques could lead to a new
generation of accurate and computationally efficient whole
root system models.
Table 1. Examples of root system models, parameters, assumptions, and applications
Model type anddescription
Main parameters Underlying assumptions Application examples Advantages anddisadvantages
Root depth models:
continuous models that give
distribution of root length
densitya
Maximum rooting depth.
Variation of root
distribution with depth
Representative volume
sufficiently big for continuity
in space and time. Empirical
root distribution with depth,
and advance of rooting front
with time
Describing root distribution with
depth.b Predicting root distribution
to model crop growth or uptakec
Good simple descriptions of root
distribution with depth. Not
explicitly related to many biological
parameters. May neglect factors
such as root clustering in
structured soils that can have
major effects on uptaked
Architectural models: discrete
models are based about an
explicit description of root
growth and branching: 1-D,
2-De, or 3-Df
For each order of
root—elongation rate;
branching interval;
branching angle; growth
tortuosity; gravitropism
Growth parameters are valid
for conditions and types of
roots simulatedg
Predicting efficacy of root
measurement methods.h Predicting
effect of gravitropism on phosphate
uptake.i Predicting effects of plant
competition on uptake.j Predicting
effect of architecture and exudates
on root uptakek
Good explicit descriptions and
visualization of root systems,
especially in 3-D. Require large
numbers of input parameters that
may be difficult to measure.
Spatial models of root growth
dynamics: continuum models
for describing the 2-D or 3-D
patterns of growth using root
length density, meristem
density
Elongation rate, and
gravitropism as coefficient
of differential equations
Representative volume is
sufficiently big for continuity
in space and time. Growth
parameters represent
average root behaviour
Describing the dynamics of
exploration of the soil domain.l
Potentially to develop a better
understanding of spatial and
temporal patterns of root growth
and development. Potentially to
bridge the gap between
architectural and continuous
function models
May provide good spatial
description of root growth, with
fewer parameters than full
architectural models. Model
parameterization requires
development, and issues such as
representative volume need further
consideration
a Gerwitz and Page (1974); Mulia and Dupraz (2006).b Gerwitz and Page (1974); Jackson et al. (1996).c Greenwood et al. (1982); Hillel and Talpaz (1975).d Passioura (1991); Tardieu et al. (1992).e Jakobsen and Dexter (1987).f Diggle (1988); Lynch et al. (1997); Pages et al. (1989); Jourdan and Rey (1997); Wu et al. (2007).g Tsegaye et al. (1995).h Bengough et al. (1992); Grabarnik et al. (1998).i Lynch and Brown (2002).j Dunbabin (2007); Rubio et al. (2002).k Dunbabin et al. (2004, 2006); Fitter and Stickland (1991).l Bastian et al. (2008); Heinen et al. (2003); Dupuy et al. (2010).
2132 | Dupuy et al. at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
A brief history of root models
Root system models were first developed in the 1970s when
Hackett and Rose (1972) proposed, for the first time,
a mechanistic model describing the extension and branching
of a cereal root system. They showed that the increase in
root length and the number of root tips for each branching
order could be derived from simple measurements of only the
number and length of root members, together with simple
assumptions of root branching interval and elongation rates.
Further work showed that these developmental analyses
could be used to model the vertical extension of the roots
and their proliferation in a uniform soil profile (Lungley,
1973; Rose, 1983). When combined with shoot developmen-
tal data, physiologically plausible, dynamic models of plant
growth were obtained (Porter et al., 1986). Simultaneously,
Gerwitz and Page (1974) analysed 107 published and un-
published data sets of root length measurements with depth
in the soil profile for many crops and found that a simple
exponential model of the change of root length density
distribution with depth described >80% of the variance in 71
of the data sets. This model was simple to use on field data,
had good predictive power, and its utility has been exploited
widely in the crop science and ecological communities
(Greenwood et al., 1982; Gale and Grigal, 1986; Jackson
et al., 1996). Since then, root density models have improved
gradually, combining developmental information and obser-
vations of growth mechanisms, and incorporating elements
of plant physiology and interaction with the environment
(Robertson et al., 1993; King et al., 2003).
The success of Gerwitz and Page’s (1974) approach led, in
turn, to a generalized description of the growth of root
systems based on diffusive processes. Page and Gerwitz (1974)
demonstrated the utility of this approach for isolated plants
(lettuce), a row crop (onions), and a sward (ryegrass). This
theoretical framework was refined further by several groups
of researchers with, for example, soil physicists developing
spatial models using reactive diffusion systems to describe the
development of the root zone (Hillel and Talpaz, 1975;
Hayhoe, 1981). Convection terms (Acock and Pachepsky,
1996) and anisotropic coefficients (de Willigen et al., 2002)
were added to incorporate tropism into the development of
the root zone, and soil heterogeneity (Heinen et al., 2003).
A review of these approaches was undertaken by Reddy
and Pachepsky (2001). In general these diffusion-type
models have proved most popular with soil physicists
studying water distribution and uptake, and have not been
used widely by plant physiologists—possibly due to the
difficulty in attributing explicit biological meaning to the
parameters.
With the advent of modern computers, computational
approaches have come to dominate the field of root
modelling. The migration was initiated by Diggle (1988) with
the ‘ROOTMAP’ simulator, shortly followed by the Pages
model for maize (Pages et al., 1989). In such models, the
architecture of the root system was made explicit, and new
types of analyses became possible. For example, Ho et al.
(2004) have employed architectural models of the common
bean root system to explore the optimum architecture to
capture phosphate; root gravitropic traits differ substantially
between genotypes and lines. Fitter et al. (1991) introduced
graph theory concepts to analyse uptake efficiency, and
fractals have been used as a tool to extract global descriptors
of root branching structures and to simulate their growth
(van Noordwijk et al., 1994). The role of architecture in the
transport of water and nutrients to plants has become moreapparent with the inclusion of water conductance models
(Doussan et al., 1998) allowing couplings between growth
and soil hydraulic processes to be revealed (Clausnitzer and
Hopmans, 1994). Models that encode the regulatory mecha-
nisms of development are now under active development
(Chavarrıa-Krauser and Schurr, 2004 ; Lucas et al., 2008). A
more extensive review of current models of root systems can
be found in Tobin et al. (2007).Currently, architectural models are the preferred ap-
proach when the development and physiology of the plant
are the main focus of study. The models can incorporate
explicitly the nature of interactions between the plant
components and its environment, and detailed predictions
of the evolution of such systems can be made. However,
root architectural models are difficult to parameterize.
Continuous root distribution models, on the other hand,can be used when only limited information about growth
can be collected. Such models are easy to parameterize, but
they do not permit plant developmental processes to be
integrated, and this has been a major limitation in the past.
The following sections show that continuous models can be
modified to incorporate root developmental processes
explicitly. This could provide new areas of application for
models of the root system.
Root systems as density distributions
Density-based models of root growth, as in many other
fields of the physics of continua, are based on two essentialconcepts. First, root distribution at a given point in space
can be prescribed in a representative volume, the minimum
volume above which stochastic effects become negligible.
This representative volume will depend upon both crop and
soil type, and may not be trivial to define. Initially it seems
reasonable to define it as a soil volume typically used in root
system measurement—perhaps between 100 cm3 and 200 cm3.
In a representative volume, therefore, various measures suchas length density, specific area, or volume fractions are good
local approximations of the state of the system. Secondly,
variations in space and time in the representative volume are
assumed to be smooth. This additional assumption allows
the determination of analytical solutions to the governing
equations (Tinker and Nye, 2000), although numerical
approaches can now address discontinuity in numerous
ways (Leveque, 2002; Heinen et al., 2003).Continuous models deal with heterogeneity by upscaling
processes and morphological properties so that the root–soil
system can be represented as a continuum. The mathemat-
ical objects that allow this to be achieved are root densities.
Root architectures are complex structures characterized by
Continuous root growth models | 2133 at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
both their geometrical and topological properties (Danjon
et al., 1999), and representing their morphologies accurately
requires the traditional definition of density to be extended.
First, densities must provide a suitable representation of the
geometry/topology duality (Fig. 1). This can be expressed by
the use of root length density (qn, cm�2) and root branching
density (qb, cm�3). Individual roots have cylindrical-like
geometries, and root length density provides good descrip-tions of the local geometrical properties of root systems to
model root–soil interactions (Bengough, 1997). The topology
of root architecture, the set of connections between roots,
can be equivalently described using branching density (Dupuy
et al., 2005c). Both root length density and root branching
density evolve with time as a result of the functioning of
root apical meristems so that the root meristem density
(qa, cm�3) is a third crucial object linking the dynamics of
root length and root branching density.
Roots also have different morphologies (e.g. inclination,
diameter, branching order) and this affects their functioning
and interaction with the environment (Waldron, 1977;
Porter et al., 1994). For densities to incorporate these
properties, the space in which densities are defined needs tobe generalized to spaces of larger than the three physical
dimensions (Grabarnik et al., 1998). For example, densities
may vary according to distance from the row, depth, the
angle of inclination of the roots, and time (respectively x, z,
a, and t). The root meristem density qa(x,z,a,t) is such thatR a2a1
R z2z1
R x2x1qaðx; z; a; tÞ:dx:dz:da represents the number of
root apices at time t, at a distance from the row between x1and x2, between depths z1 and z2, and whose growthdirection is contained between a1 and a2.
Theoretical basis for modelling the dynamicsof root distribution in space
Continuous models have seen little conceptual development
since diffusion models were first developed. In particular,
continuous models have failed to propose theories equiva-
lent to those underlying architectural-based models,
whereby explicit measurable and meaningful individual root
developmental processes, for example branching angle, root
elongation rate, or gravitropic rate, are encoded in themathematics of the model. This makes it more difficult to
envisage the physical basis of the model, and may deter
many biologists from using this approach. In the next
sections, modifications to continuous models that make the
description of developmental processes more explicit are
described. It may be useful conceptually to consider growth
of parts of a root system in known representative volumes
of soil. Root system growth is simply the sum of rootproduction rate, for example elongation, cessation, or
death, generated by each root meristem. The number of
meristems present increases by the birth of new lateral roots
and by new roots entering a particular soil volume, and
decreases by root death and roots leaving that soil volume.
Conservation law for meristem dynamics
Conservation laws are mathematical relationships describ-
ing the variations of properties such as meristem density,
which remains constant within the system in the absence of
external sources or sinks (e.g. branching or meristem
mortality). In the case of root densities, such equations
need to be defined in terms of both the physical coordinates,
for example x and z, and the angles defining the direction ofgrowth, for example a. In such generalized domains, the
conservation of meristems can be written as:
@qa@t
þ @qag@a
þ @qaecosa@x
þ @qaesina@z
¼ b� d ð1Þ
This equation shows that the number of meristems
defined at a given time t, whose spatial coordinates are
Fig. 1. Representing the geometry–topology duality using density
distributions. Branching density is required to encode the topology
of the root system. In a previous study (Dupuy et al., 2005c) the
root systems of maritime pine were digitized (A) and maps of the
root systems branching density were extracted (B). Assuming that
roots were growing as straight lines, it was possible to reconstruct
realistic root architecture using solely the branching densities (C).
Because roots have more complex trajectories and have no pre-
defined length, the complete root system needs both root length
and root branching density to characterize its structure fully.
Adapted, with permission, from Dupuy et al. (2005c).
2134 | Dupuy et al. at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
(x,z) and angle a, is changing with time (@/@t) as a result of
meristems moving in and out of the neighbourhood (@/@x,@/@y) via elongation rate e and according to the direction of
growth (cos a, sin a) but also due to reorientation (@/@a) ofmeristems as a result of gravitropism g (� d�1) and the
creation of new meristems due to branching b (cm�3 d�1)
and loss of mersitem caused by mortality d (Dupuy et al.,
2009). Generalizing the space in which root distribution isdefined allows the root developmental parameters to appear
explicitly in the same equation. e represents the elongation
rate (cm d�1) of individual root meristems, g is the re-
orientation rate (change of root angle per unit time, � d�1).
b is the number of new meristems created per unit volume
(cm�3 d�1) and can be expressed as a function of individual
root branching rate (cf. following paragraph), and d is the
number of meristems being destroyed per unit of time (cm�3
d�1).
Branching and root length densities result in turn from
the sum of the activity of meristems with time (e.g. see
Fig. 2). This can be expressed mathematically as:
@qb@t
¼ b ð2Þ
@qn@t
¼ eqa ð3Þ
Equation 2 states that the number of roots being created
per unit of time in a unit volume (volumetric branching
rate, b) equals the change with time of the branching
density. In the next section, we describe how b can be
expressed as a function of root meristem density to describe
lateral root initiation. Equation 3 states that the increase in
root length per unit of time in a unit volume equals the
meristem density multiplied by the elongation rate (e).
Equations 1–3 together describe the dynamics of the root
architecture.
An interesting property of root architectures can befound by integrating Equation 1 with time: if the gravi-
tropic term is small in comparison with other terms, then
branching density is the sum of the derivative of the root
length density in the direction u (cos a, sin a) and meristem
density.
qb � qa þ@qn@u
ð4Þ
Equation 1 is a classic conservation law, similar to those
found in many areas of physical science, and is hyperbolicin nature. The solutions form travelling waves of meristem
density. The root length density and branching densities,
determined as the integration with time of the meristematic
activity (Equations 2 and 3), result from the morphology of
this wave. This is illustrated in the following section.
Modelling root system development
The coefficients e, g, and b defined in Equation 1 are needed
to describe the complexity of root development. These
coefficients are therefore generally not constant but more
Fig. 2. Root system dynamics in a known volume. (A and B) Region of a barley root system growing in a thin transparent container,
imaged 2.3 h apart (root diameter is ;0.4 mm). (C) Change in root length between A and B. (D and E) Idealized representation of root
growth showing new root length associated with existing meristems, newly initiated meristems, and meristems that have grown into the
rooting volume from immediately above.
Continuous root growth models | 2135 at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
sophisticated growth functions. For example, the coefficient
g describing gravitropism is needed to describe root
responses that are proportional to the angle from the
vertical. One possible function for that could be (Dupuy
et al., 2009):
g¼g0ðaþp=2Þ ð5Þ
Lateral root initiation at a branching rate b0 and at
a branching angle of p/2 can be modelled as a source term
of the form:
b¼b0½qaðx; z; aþp=2; tÞþqaðx; z; a�p=2; tÞ�=2 ð6Þ
In this equation, the creation of new roots of a given
growth direction u is due to roots having inclination angle
a+p/2 and branching on the right in addition to the roots
having inclination angle a–p/2 and branching on the left
(Fig. 3A). Following the same reasoning, dichotomous
branching would require a different form of source term:
b¼b0½qaðx; z; aþp=6; tÞþqaðx; z; a�p=6; tÞ�=2�qaðx; z; aÞð7Þ
The first part of the mathematical expression here is
similar to a herringbone type, with a branching angle of p/6,and the second part of the expression incorporates the death
of the mother root at branching (Fig. 3B). More complex
systems of equations can also be derived from Equation 1,
in order to simulate the behaviour of different types of
roots. Modelling the dynamics of roots of different branch-ing orders is a good example. It is known, for example, that
roots of higher branching orders grow more slowly and are
less gravitropic (Drew, 1975; Rose, 1983). This can be
represented using a system of equations of the type:
Fig. 3. Modelling development with continuous models. Branching processes can be modelled through the source term b in Equation 1.
In herringbone types (A), the quantity of roots produced at angle a is half the branching rate of roots with angle a+p/2 (those branching
on the right) plus half the branching rate of roots with angle a–p/2 (those branching on the left). Dichotomous types can be modelled by
incorporating the mortality of mother roots (B). Here bd represents the branching rate density with contours representing their distribution
at 3, 6, and 9d (respectively blue, green, and red). ld is the root length density and the red line schematizes the equivalent structure.
These models shows that the fundamental difference between herringbone and dichotomous root systems is the spatial pattern of
branching activity in meristems. Architectural responses to environmental factors can be represented through spatially varying
coefficients (C). The meristem density distribution (md) and root length density distribution are shown resulting from an increase in
elongation rate and branching rate in a patch of nutrient. Simulations in (A–C) were carried out using a finite volume upwind scheme. All
simulations were initiated in the top left corner of the soil volume (x,z¼0,0). The densities of root meristems in boundary elements outside
the simulated soil volume were defined as the mirror image of the boundary elements inside the soil volume to preserve a vertical plane of
symmetry. Fluxes on the remaining faces are prevented. See Dupuy et al. (2009) for more details on the numerical approach developed.
2136 | Dupuy et al. at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
@qia@t
þ @qiagi@a
þ @qiaeicosa@x
þ @qiaeisina@z
¼ bi; i 2 f1; ng ð8Þ
Here, i represents the branching order, and the branchingterm bi is a function of the root meristem density at
branching order i–1, for example Equations 6 and 7. The
branching term bi therefore couples the different equations
together. The system allows roots of different types to have
a distinct elongation rate (ei) and gravitropic rate (gi).
Root mortality is also an important process in the
development and growth of root systems, with total root
production exceeding the final root system produced by upto 80% (Steingrobe et al., 2001). Current methods for
estimating root mortality using rhizotron or ingrowth core
techniques have not allowed the mechanisms of mortality to
be identified (Smit and Zuin, 1996; Steingrobe et al., 2001).
The simplest way to integrate mortality is to assume that
the mortality of root tips is uniformly spread throughout
the root system (Robertson et al., 1993), and to decrease
root length density accordingly:
@qn@t
¼ eqa � d�qn ð9Þ
Mortality can also vary significantly between roots of
different classes or according to soil conditions (Wells and
Eissenstat, 2001). More sophisticated models have yet to be
developed to account for the complexity of the processes of
root mortality where, for instance, a whole root axis dies
together with its associated lateral branches.
Feedbacks with the environment
The models presented so far assume uniform and steady
growth conditions. To incorporate the effects of 3-D soilheterogeneity into such a model, the coefficients e, g, and
b must be considered as a function of local environmental
variables h (Fig. 3C), such as volumetric water content, soil
strength, or solute concentration:
@qa@t
þ @qagðhÞ@a
þ @qaeðhÞcosa@x
þ @qaeðhÞsina@z
¼ bðhÞ � dðhÞ
ð10Þ
Environmental variables are also modified as a result of
growth processes and this has been the subject of numerousstudies, for example radial uptake of water by roots
(Gardner, 1960). Upscaling mechanisms from root to root
systems, however, can be considerably more difficult than
simply summing up individual effects (Baldwin et al., 1972).
In the past, this has been approached through generic
functions incorporating group effects, for example the
concept of volume fraction available for uptake by roots
(King et al., 2003).When root environmental processes occur at larger scales,
for example water depletion in root systems (Garrigues
et al., 2006), growth must be coupled to additional equa-
tions such as transport of water and nutrients (Mmolawa
and Dani, 2000). Such a system of equations needs more
sophisticated numerical methods to be developed (Hayhoe,
1981). Alternatively, it is simpler to assume that growth is
slow in comparison with the transport of mobile nutrients
and therefore to implement density-dependent growth of
root meristems (Brugge, 1985; Chen and Lieth, 1993).
Bastian et al. (2008), for example, proposed a simplified
function of the form:
e¼kðqmaxn �qnÞ ð11Þ
In these cases the velocity and branching are linked
directly to root length density through an arbitrary
parameter (k) that requires experimental characterization.
Calibration of density-based models
The choice of a model is often motivated by the nature of
the data that can be collected from experimental systems.
For that reason, density-based models have been particu-
larly popular for root studies. This is due to the fact that
density can be estimated from excavated cores of soil, for
which numerous experimental techniques are available (doRosario et al., 2000; van Noordwijk et al., 2000). For
example, exponential models can be fitted to vertical cores
of soil obtained by auger sampling (Barraclough and Leigh,
1984) and 2-D diffusion models have used pinboard
techniques to obtain mappings of root length distribution
in space (Heinen et al., 2003). Considerable progress has
been made in the measurements of individual root growth
parameters (Lecompte and Pages, 2007; Danjon andReubens, 2008; Pages et al., 2009). However, parameterizing
architectural models is still difficult and often requires
partial excavation of root systems.
In the case of the models presented herein, spatial
distributions of both root length density and meristem
density are required. These data may be estimated from
rhizotron or minirhizotron window observations (e.g. Fig.
2), or from washed root samples augered from the soil asmentioned above. Image analysis programs such as Winrhizo
(Regent Instruments Inc.) can provide the number of root
ends in a washed root sample, although in samples where
there are many broken roots this would substantially
overestimate meristem density. One possible approach
would be to develop an image analysis procedure that
combined root end estimation with the use of stains for
identifying dividing cells in the meristem. To verify anycalibration or parameterization procedure it would be
possible to simulate the procedure using explict 3-D
architecture models, for example by simulating the mea-
surement of root length density and meristem density in
particular regions of the root system, and determining
whether this gave appropriate parameters for the continu-
ous model. For example, if the gravitropism term was
relatively small, this may facilitate parameterization of themodel (e.g. Equation 4).
Partial differential equation (PDE) models (Equation 1)
usually need to be solved using numerical approaches and
the estimation of model parameters requires specific proce-
dures to be developed. This has been solved in the past by
Continuous root growth models | 2137 at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
using non-linear optimization algorithms (Heinen et al.,
2003). With this approach, the model is run using different
sets of parameter values incorporating small variations in
each parameter. A minimization criterion is used to measure
the difference between the observed root length density and
the predicted root length density. The model parameters are
then updated in the direction of the greatest rate of decrease
of the criterion. These steps are repeated until the criterionno longer decreases. This method has been further refined to
allow derivation of the standard error of the model param-
eters and to test their significance (Acock and Pachepsky,
1996). This allows model complexity to be explored.
Models that are constructed on mechanistic principles,
such as in Equation 1, define root developmental processes
explicitly. Therefore, the optimization of their parameters,
e.g. g0 (Equation 5), b0 (Equations 6 and 7), or ei (Equation8), could prove particularly useful in the future to ‘invert’
root developmental programs and to reveal root growth
parameters from the measurement of root length density
patterns. Individual root properties are not required as
inputs for density-based models. More parameters are
required for more complex models, and to account for
environmental influences on root growth—this will also
affect the uncertainties associated with estimates of theseparameters. The complexity of any given model must be in
balance with the amount and quality of data on root
distribution in relation to soil properties.
Root system development as waves ofmeristems
In previous work (Dupuy et al., 2009), it was shown using
simple growth functions that a one-dimensional analytical
approximation of Equation 1 can be obtained, using the
method of characteristics (Leveque, 2002):
qa¼ð1þbtÞN0
�z�et�
ea2�e�2gt � 1
�4g
; a
�ð12Þ
The function N0(z,a) is the initial root meristem distribu-
tion which is assumed to be the normal distribution. Theexpression z–et in this model therefore implies that root
systems tend to propagate and amplify the initial peak of
meristem density, and that such spatial peaks travel
approximately at the speed of the average root elongation
rate e.
Developmental processes, through the coefficients e, g,
and b (Equations 1 and 8), have a strong influence on the
propagation of the waves of meristems and the resultingshapes of root systems. The branching patterns (e.g. angle,
frequency and mortality) determine the topology, for
example herringbone and dichotomous architectures in Fig.
3A, B. Systems possessing roots of several branching orders
growing at different rates produce the superposition of
waves of meristems of different branching order (Fig. 4).
This superposition of waves is largely the reason for the
exponential profiles of root length density distribution withdepth (Gerwitz and Page, 1974). Root–soil interactions can
also influence root developmental properties locally and can
generate irregular patterns of root length distribution in
space (Fig. 3C).
From a biological perspective, the waves of meristems
are due to the polarity of the development of root
apical meristems. Newly created tissues from a single meri-
stem expand and become increasingly rigid during
Fig. 4. The development of root systems represented as waves of meristems. Preliminary experimental results indicate that meristem
density may form travelling peaks of activity. A time lapse minirhizotron experiment (Dupuy et al., 2009) carried out on root systems of
barley seedlings was used to build spatial maps of root length density (A) and the equivalent meristem activity in (B) (volumetric root
length production rate, cm cm�3 d�1) obtained from Equation 3. More complex root system morphologies can be modelled using the
superposition of meristematic waves of different types. For example, roots of a certain branching order have an average elongation rate
and form travelling peaks of meristem density during the growth of the root system. Because root systems have roots of different
branching order (here first order in blue long dashed–dotted lines and second order in green dashed lines), the system can be modelled
as the superposition of travelling waves of different velocities (C).
2138 | Dupuy et al. at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
differentiation. During this process, root tips are pushed by
expanding cells and tend to travel away from their branching
point. At the scale of the root system, this mechanism induces
travelling spatial peaks of meristem density, and the mathe-
matical evidence for this lies in the nature of Equation 1.
The concept of a wave of meristems still needs to be
tested experimentally. Similar behaviour has been observed
in the propagation of the soil drying front (Davies et al.,
1990). In particular, it was found that spatial maxima in the
water extraction rate were formed and propagated through
the soil. More recently, Garrigues et al. (2006) have used X-
ray imaging and computer simulation to characterize in
great detail the formation and propagation of these spatial
maxima when following water uptake by narrow leaf lupin
(Doussan et al., 2006; Garrigues et al., 2006). Theoretical
studies also suggest that resource-dependent growth of
branching systems, such as blood vessels, fungal mycelia, or
hairy roots, induces the formation of waves of tip activity in
response to chemical signals (Anderson and Chaplain, 1998;
Boswell et al., 2003; Bastian et al., 2008).
Regions of high density of root apical meristems are of
particular importance for the uptake of water and
nutrients, and raise many interesting questions. Depletion
around the root increases as a function of the distance
from the tip (Claassen and Barber, 1974), while living
root hairs are distributed in the proximity of root tips
(Gilroy and Jones, 2000) and sensing for environmental
signals has been shown to be located in the root tips
(Zhang and Forde, 1998; Wilkinson and Davies, 2002;
White et al., 2005; Hodge, 2009). Are water extraction
fronts the result of travelling maxima of meristematic
activity? Have plants evolved to proliferate their root
meristems in regions less likely to be depleted? These
questions remain unresolved at this stage. Further studies
are required to characterize the nature of the dynamics of
meristems in soil.
Deformable domains, the next generation ofdensity-based models?
It is not always possible to obtain analytical solutions from
the equations developed in previous sections. In such cases,
numerical approaches must be developed to generate
approximate solutions. Numerical solutions are obtained
by computing the values of the root densities on a mesh,
using finite element, or finite volume methods (Zienkiewicz
and Taylor, 1998; Leveque, 2002).
Traditionally, fixed meshes were used to model soil and
to assign root length density values to elements of the mesh
(referred to as control volumes or elements). The simulation
uses the system variables, defined at control volumes, to
compute fluxes of roots entering and leaving the control
volume. The dynamics of root distribution are then built
incrementally from the initial conditions of the system
(Heinen et al., 2003). These ‘soil-centred’ approaches,
referred to as Eulerian, are appropriate when tracking of
the motion of matter is not required or achievable, for
example fluid dynamics (Ashgriz and Mostaghimi, 2002).
A promising new type of ‘plant-centred’ model could be
constructed using deformable meshes that assign control
volumes to domains of moving meristems and growing root
systems (Fig. 5A). Such models are referred to in physics as
Lagrangian and are used commonly in the field of solid
mechanics (Zienkiewicz and Taylor, 1998). Deformable
plant models could have a range of new applications for
models of root systems. Deformable models represent
explicitly the domain occupied by the root system of
a plant and are naturally adapted to model interactions
with other plants. Algorithms inspired from the field of
contact modelling (Dupuy et al., 2005b) can be developed
to identify neighbouring plants and implement density-
dependent growth (Fig. 5B). Such models could have
application in ecology to improve individual-based,
Fig. 5. Continuous deformable domains to represent the development of the root system. (A) Schematic representation of the principle
of continuous deformable domains: the distribution of meristems in the root system is encoded in a discretized domain (blue). The
growth equations (Equations 1–3) are then used to deform the domain and update the meristem distribution in the domain and the root
length distribution (red). (B) Population-based simulation of the model of barley seedlings developed by Dupuy et al. (2009) and
implementing a density-dependent growth of meristems. Initial conditions consisted of 81 seeds placed randomly on the plane and
initiation of root–root interaction. The mesh shows the deformed domains resulting from the growth of interacting root systems.
Continuous root growth models | 2139 at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
population models (Begg et al., 2006). Also, deformable
meshes use less control volumes: 16 subdivisions have been
used to decompose the deformable domains as opposed to
40 in Dupuy et al. (2009). Deformable models would have
numerous applications where ensembles of root systems
need to be simulated efficiently.
Concluding remarks
Models have become important tools for characterizing the
growth and functioning of plants. Pioneering work in the
1970s used continuous approaches to model root systemgrowth. However, models of this type are based on spatial
mathematics that often requires significant expertise in
modelling. Perhaps for this reason, root architectural
models, where the shape and structure of roots are
represented explicitly within simulation programs, have
now come to dominate the field (Godin and Sinoquet,
2005). Based on the simulation of an ensemble of identical
modules, i.e. virtual meristems, complex mechanisms caneasily be encoded as morphogenetic rules and emergent
properties can be simulated using existing platforms such as
L-studio, AMAP, or Lignum (Pertunen et al., 1996; Allen
et al., 2005; Barczi et al., 2008). New algorithms are being
developed to improve the efficiency of computation (Renton
et al., 2005; Cournede et al., 2006) and facilitate the
estimation of model parameters (Letort et al., 2008).
Existing models are so sophisticated that one may evenquestion the need for alternative forms of models.
Mathematical approaches, however, offer compelling
alternatives to architectural models. Analytical models,
exact or approximate solutions to growth equations, can be
obtained in the form of mathematical functions. These
functions provide insight into the development of the root
system as a whole, and can be parameterized relatively
easily from field data. The use of continuous variables toaggregate root morphological properties also facilitates the
coupling of growth with environmental and physical pro-
cesses, which is useful for a wide range of plant environ-
mental studies (Watanabe et al., 2004). For example, soil
mechanics and transport models use partial differential
equations to describe soil mechanics, and diffusion and
mass flow in soil (Bengough, 1997; van Beek et al., 2005;
Doussan et al., 2006), and such models would communicatemore naturally with continuous descriptions of plant root
structure. Finally, representing root systems as continua
allows more efficient computational models to be devel-
oped. The continuous deformable domain approach pre-
sented herein is a first step toward a next generation of
continuous root system models. The principle is likely to
have applications when the growth of numerous plants
needs to be simulated. For example, most individual-basedpopulation models involve the definition of the plant’s
zone of influence, where acquisition or competition for
resources occurs (Begg et al., 2006). These zones have been
represented classically as domains contained in envelopes
(Cescatti, 1997). The growth of these regions and the
evolution of their structure have generally been imple-
mented through empirical, arbitrary functions (Weiner and
Damgaard, 2006) and could be replaced by more accurate
mechanistic principles.
Continuous-based approaches for modelling root system
growth are still at an early stage of development. Such
models could have a wide range of applications, particularly
where classic architectural models cannot provide predic-tions due to computational limitations, for example pop-
ulation models or breeding scenarios, or due to the
difficulty in calibrating the models efficiently. There are
still, however, considerable challenges to develop suitable
analytical methods and to optimize the numerical analysis
of such systems, notably for the meshing and approxima-
tion of solutions on deformable domains. Fortunately, there
is abundant literature on the subject that indicates thepotential for improvements.
Acknowledgements
The Scottish Crop Research Institute receives financial
support from the Scottish Government Rural and Environ-
ment Research and Analysis Directorate (RERAD, Work-
package 1.7).
References
Acock B, Pachepsky Y. 1996. Convective–diffusive model of two-
dimensional root growth and proliferation. Plant and Soil 180,
231–240.
Allen MT, Prusinkiewicz P, Dejong TM. 2005. Using L-systems
for modeling source–sink interactions, architecture and physiology
of growing trees: the L-PEACH model. New Phytologist 166,
869–880.
Anderson A, Chaplain M. 1998. Continuous and discrete
mathematical models of tumor-induced angiogenesis. Bulletin of
Mathematical Biology 60, 857–899.
Ashgriz N, Mostaghimi J. 2002. An introduction to computational
fluid dynamics. In: Saleh J, ed. Fluid flow handbook. McGraw-Hill
Professional, Chapter 24.
Baldwin J, Tinker P, Nye P. 1972. Uptake of solutes by multiple root
systems from soil. Plant and Soil 36, 693–708.
Barczi J-F, Rey H, Caraglio Y, de Reffye P, Barthelemy D,
Dong QX, Fourcaud T. 2008. AmapSim: a structural whole-plant
simulator based on botanical knowledge and designed to host external
functional models. Annals of Botany 101, 1125–1138.
Barraclough PB, Leigh RA. 1984. The growth and activity of winter
wheat roots in the field: the effect of sowing date and soil type on root
growth of high-yielding crops. Journal of Agricultural Science 103,
59–74.
Bastian P, Chavarria-Krauser A, Engwer C, Jaeger W,
Marnach S. 2008. Modelling in vitro growth of dense root networks.
Journal of Theoretical Biology 254, 99–109.
2140 | Dupuy et al. at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
Begg GS, Hockaday S, McNicol JW, Askew M, Squire GR. 2006.
Modelling the persistence of volunteer oilseed rape (Brassica napus).
Ecological Modelling 198, 195–207.
Bengough AG. 1997. Modelling rooting depth and soil strength in
a drying soil profile. Journal of Theoretical Biology 186, 327–338.
Bengough AG, Mackenzie CJ, Diggle AJ. 1992. Relations between
root length densities and root intersections with horizontal and vertical
planes using root growth modeling in 3-dimensions. Plant and Soil
145, 245–252.
Birch CJ, Andrieu B, Fournier C, Vos J, Room P. 2003. Modelling
kinetics of plant canopy architecture—concepts and applications.
European Journal of Agronomy 19, 519–533.
Boswell G, Jacobs H, Davidson F, Gadd G, Ritz K. 2003. Growth
and function of fungal mycelia in heterogeneous environments. Bulletin
of Mathematical Biology 65, 447–477.
Brugge R. 1985. A mechanistic model of grass root growth and
development dependent upon photosynthesis and nitrogen uptake.
Journal of Theoretical Biology 116, 443–467.
Cescatti A. 1997. Modelling the radiative transfer in discontinuous
canopies of asymmetric crowns. I. Model structure and algorithms.
Ecological Modelling 101, 263–274.
Chavarrıa-Krauser A, Schurr U. 2004. A cellular growth model for
root tips. Journal of Theoretical Biology 230, 21–32.
Chen D-X, Lieth JH. 1993. A two-dimensional, dynamic model for
root growth distribution of potted plants. Journal of the American
Society for Horticultural Science 118, 181–187.
Claassen N, Barber SA. 1974. A method for characterizing the
relation between nutrient concentration and flux into roots of intact
plants. Plant Physiology 54, 564–568.
Clausnitzer V, Hopmans J. 1994. Simultaneous modeling of
transient three-dimensional root growth and soil water flow. Plant and
Soil 164, 299–314.
Cournede P-H, Kang M-Z, Mathieu A, Barczi J-F, Yan H-P,
Hu B-G, de Reffye P. 2006. Structural factorization of plants to
compute their functional and architectural growth. Simulation 82,
427–438.
Danjon F, Bert D, Godin C, Trichet P. 1999. Structural root
architecture of 5-year-old Pinus pinaster measured by 3D
digitising and analysed with AMAPmod. Plant and Soil 217,
49–63.
Danjon F, Reubens B. 2008. Assessing and analyzing 3D
architecture of woody root systems, a review of methods and
applications in tree and soil stability, resource acquisition and
allocation. Plant and Soil 303, 1–34.
Davies WJ, Mansfield TA, Hetherington AM. 1990. Sensing of soil
water status and the regulation of plant growth and development.
Plant, Cell and Environment 13, 709–719.
de Dorlodot S, Forster B, Pages L, Price A, Tuberose R,
Draye X. 2007. Root system architecture: opportunities and
constraints for genetic improvement of crops. Trends in Plant Science
12, 474–481.
de Willigen P, Heinen M, Mollier A, Van Noordwijk M. 2002. Two-
dimensional growth of a root system modelled as a diffusion process.
I. Analytical solutions. Plant and Soil 240, 225–234.
Diggle AJ. 1988. ROOTMAP—a model in three-dimensional
coordinates of the growth and structure of fibrous root systems. Plant
and Soil 105, 169–178.
do Rosario M, Oliveira G, Van Noordwijk M, Gaze SR,
Brouwer G, Bona S, Mosca G, Hairiah K. 2000. Auger sampling,
ingrowth cores and pinboard methods. In: Smit A, Bengough A,
Engels C, Van Noordwijk M, Pellerin S, Van de Geijn S, eds. Root
methods: a handbook. Berlin: Springer, 175–210.
Doussan C, Pages L, Vercambre G. 1998. Modelling of the
hydraulic architecture of root systems: an integrated approach
to water absorption—model description. Annals of Botany 81,
213–223.
Doussan C, Pierret A, Garrigues E, Pages L. 2006. Water uptake
by plant roots: 2. Modelling of water transfer in the soil root-system
with explicit account of flow within the root system—comparison with
experiments. Plant and Soil 283, 99–117.
Drew DA. 1975. Comparison of the effects of a localized supply of
phosphate, nitrate, ammonium and potassium on the growth of the
seminal root system, and the shoot, in barley. New Phytologist 75,
479–490.
Drouet J-L, Pages L. 2007. GRAAL-CN: a model of GRowth,
Architecture and ALlocation for Carbon and Nitrogen dynamics within
whole plants formalised at the organ level. Ecological Modelling 206,
231–249.
Dunbabin V. 2007. Simulating the role of rooting traits in crop–weed
competition. Field Crops Research 104, 44–51.
Dunbabin VM, Diggle AJ, Rengel Z, van Hugten R. 2002.
Modelling the interactions between water and nutrient uptake and root
growth. Plant and Soil 239, 19–38.
Dunbabin VM, McDermott S, Bengough AG. 2006. Upscaling from
rhizosphere to whole root system: modelling the effects of phospholipid
surfactants on water and nutrient uptake. Plant and Soil 283, 57–72.
Dunbabin V, Rengel Z, Diggle A. 2004. Simulating form and
function of root systems: efficiency of nitrate uptake is dependent on
root system architecture and the spatial and temporal variability of
nitrate supply. Functional Ecology 18, 2004–2211.
Dupuy L, Fourcaud T, Stokes A. 2005a. A numerical investigation
into the influence of soil type and root architecture on tree anchorage.
Plant and Soil 278, 119–134.
Dupuy L, Fourcaud T, Stokes A. 2005b. A numerical investigation
into factors affecting the anchorage of roots in tension. European
Journal of Soil Science 56, 319–327.
Dupuy L, Fourcaud T, Stokes A, Danjon F. 2005c. A density based
approach for the modelling of root architecture: application to Maritime
pine (Pinus pinaster Ait.) root system. Journal of Theoretical Biology
236, 323–334.
Dupuy L, Vignes M, McKenzie B, White P. 2010. The dynamics of
root meristem distribution in soil. Plant, Cell and Environment 33,
358–369.
Fitter AH, Stickland TR. 1991. Architectural analysis of plant root
systems 2. Influence of nutrient supply on architecture in contrasting
plant species. New Phytologist 118, 383–389.
Continuous root growth models | 2141 at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
Fitter AH, Stickland TR, Harvey ML, Wilson GW. 1991.
Architectural analysis of plant root systems 1. Architectural correlates
of exploitation efficiency. New Phytologist 118, 375–382.
Fourcaud T, Lac P, Castera P, de Reffye P. 2003. Numerical
modelling of shape regulation and growth stresses in trees. II.
Implementation in the AMAPpara software. Trees-Structure and
Function 17, 31–39.
Gale MR, Grigal DF. 1986. Vertical root distribution of northern tree
species in relation to successional status. Canadian Journal of Forest
Research 17, 829–834.
Gardner WR. 1960. Dynamic aspects of water availability to plants.
Soil Science 89, 63–73.
Garrigues E, Doussan C, Pierret A. 2006. Water uptake by plant
roots: 1 Formation and propagation of a water extraction front in
mature root systems as evidenced by 2D light transmission imaging.
Plant and Soil 283, 83–98.
Gerwitz A, Page ER. 1974. An empirical mathematical model to
describe plant root systems. Journal of Applied Ecology 11,
773–781.
Gilroy S, Jones DL. 2000. Through form to function: root hair
development and nutrient uptake. Trends in Plant Science 5,
56–60.
Godin C, Sinoquet H. 2005. Functional–structural plant modelling.
New Phytologist 166, 705–708.
Grabarnik P, Pages L, Bengough AG. 1998. Geometrical properties
of simulated maize root systems: consequences for length density and
intersection density. Plant and Soil 200, 157–167.
Greenwood D, Gerwitz A, Stone D, Barnes A. 1982. Root
development of vegetable crops. Plant and Soil 68, 75–96.
Grimm V. 1999. Ten years of individual-based modelling in ecology:
what have we learned and what could we learn in the future?
Ecological Modelling 115, 129–148.
Hackett C, Rose D. 1972. A model of the extension and branching of
a seminal root of barley, and its use in studying relations between root
dimensions. 1. The model. Australian Journal of Biological Science 25,
669–679.
Hammer G, Cooper M, Tardieu F, Welch S, Walsh B, van
Eeuwijk F, Chapman S, Podlich D. 2006. Models for navigating
biological complexity in breeding improved crop plants. Trends in Plant
Science 11, 587–593.
Hayhoe H. 1981. Analysis of a diffusion model for plant root growth
and an application to plant soil-water uptake. Plant and Soil 131,
334–343.
Heinen M, Mollier A, De Willigen P. 2003. Growth of a root system
described as diffusion. II. Numerical model and application. Plant and
Soil 252, 251–265.
Hillel D, Talpaz H. 1975. Simulation of root growth and its effect on
the pattern of soil water uptake by a nonuniform root system. Soil
Science 121, 307–312.
Ho MD, McCannon BC, Lynch JP. 2004. Optimization modeling of
plant root architecture for water and phosphorus acquisition. Journal
of Theoretical Biology 226, 331–340.
Hodge A. 2009. Root decisions. Plant, Cell and Environment 32,
628–640.
Hodge A, Robinson D, Griffiths BS, Fitter AH. 1999. Why plants
bother: root proliferation results in increased nitrogen capture from an
organic patch when two grasses compete. Plant, Cell and
Environment 22, 811–820.
Jackson RB, Canadell J, Ehleringer JR, Mooney HA, Sala OE,
Schulze ED. 1996. A global analysis of root distributions for terrestrial
biomes. Oecologia 108, 389–411.
Jakobsen BF, Dexter AR. 1987. Effect of soil structure on wheat
root-growth, water-uptake and grain-yield—a computer-simulation
model. Soil and Tillage Research 10, 331–345.
Jourdan C, Rey H. 1997. Modelling and simulation of the architecture
and development of the oil-palm (Elaeis guineensis Jacq.) root system.
Plant and Soil 190, 217–233.
King J, Gay A, Sylvester-Bradley R, Bingham I, Foulkes J,
Gregory P, Robinson D. 2003. Modelling cereal root systems for
water and nitrogen capture: towards an economic optimum. Annals of
Botany 91, 383–390.
Lecompte F, Pages L. 2007. Apical diameter and branching density
affect lateral root elongation rates in banana. Environmental and
Experimental Botany 59, 243–251.
Letort V, Mahe P, Cournede P-H, de Reffye P, Courtois B. 2008.
Quantitative genetics and functional–structural plant growth models:
simulation of quantitative trait loci detection for model parameters and
application to potential yield optimization. Annals of Botany 101,
1243–1254.
Leveque RJ. 2002. Finite volume methods for hyperbolic problems.
Cambridge: Cambridge University Press.
Lucas M, Guedon Y, Jay-Allemand C, Godin C, Laplaze L. 2008.
An auxin transport-based model of root branching in Arabidopsis
thaliana. PLoS ONE 3, e3673.
Lungley DR. 1973. The growth of root systems—a numerical
computer simulation model. Plant and Soil 38, 145–159.
Lynch J. 2007. Roots of the second green revolution. Australian
Journal of Botany 55, 493–512.
Lynch J, Brown KM. 2002. Topsoil foraging—an architectural
adaptation of plants to low phosphorus availability. Plant and Soil 237,
225–237.
Lynch J, Nielsen KL, Davis RD, Jablokow AG. 1997. Simroot:
modelling and visualization of root systems. Plant and Soil 188,
139–151.
Mmolawa K, Dani O. 2000. Root zone solute dynamics under drip
irrigation: a review. Plant and Soil 222, 163–190.
Mulia R, Dupraz C. 2006. Unusual fine root distributions of
two deciduous tree species in southern France: what
consequences for modelling of tree root dynamics? Plant and Soil
281, 71–85.
Page ER, Gerwitz A. 1974. Mathematical models, based on diffusion
equations, to describe root systems of isolated plants, row crops, and
swards. Plant and Soil 41, 243–254.
2142 | Dupuy et al. at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from
Pages L, Jordan M, Picard D. 1989. A simulation model of the
three-dimensional architecture of the maize root system. Plant and Soil
119, 147–154.
Pages L, Serra V, Draye X, Doussan C, Pierret A. 2009.
Estimating root elongation rates from morphological measurements of
the root tip. Plant and Soil, in press.
Passioura JB. 1991. Soil structure and plant-growth. Australian
Journal of Soil Research 29, 717–728.
Pertunen J, Sievanen R, Nikimaa E, Salminen H, Saarenmaa H,
Vakev AJ. 1996. LIGNUM: a tree model based on simple structural
units. Annals of Botany 77, 87–98.
Podlich D, Cooper M. 1998. QU-GENE: a simulation platform for
quantitative analysis of genetic models. Bioinformatics 14, 632–653.
Porter JR, Klepper B, Belford RK. 1986. A model (WHTROOT)
which synchronizes root growth and development with shoot
development. Plant and Soil 92, 133–145.
Porter PS, Comerford NB, Escamilla JA. 1994. Use of Theissen
areas in models of nutrient uptake in forested ecosystems. Soil
Science Society of America Journal 58, 210–215.
Prusinkiewicz P. 2004. Modeling plant growth and development.
Current Opinion in Plant Biology 7, 79–83.
Reddy VR, Pachepsky YA. 2001. Testing a convective–dispersive
model of two-dimensional root growth and proliferation in
a greenhouse experiment with maize plants. Annals of Botany 87,
759–768.
Renton M, Hanan J, Burrage K. 2005. Using the canonical
modelling approach to simplify the simulation of function in functional–
structural plant models. New Phytologist 166, 845–857.
Robertson MJ, Fukai S, Hammer GL, Ludlow MM. 1993.
Modelling root growth of grain sorghum using the CERES approach.
Field Crops Research 33, 113–130.
Rose D. 1983. The description of the growth of root systems. Plant
and Soil 75, 405–415.
Rubio G, Walk TC, Ge Z, Yan XL, Liao H, Lynch J. 2002. Root
gravitropism and below-ground competition among neighbouring
plants: a modelling approach. Annals of Botany 88, 929–940.
Smit A, Zuin A. 1996. Root growth dynamics of Brussels sprouts
(Brassica olearacea var. gemmifera) and leeks (Allium porrum L.) as
reflected by root length, root colour and UV fluorescence. Plant and
Soil 185, 271–280.
Somma F, Hopmans JW, Clausnitzer V. 1998. Transient three-
dimensional modeling of soil water and solute transport with
simultaneous root growth, root water and nutrient uptake. Plant and
Soil 202, 281–293.
Steingrobe B, Schmid H, Claassen N. 2001. Root production and
root mortality of winter barley and its implication with regard to
phosphate acquisition. Plant and Soil 237, 239–248.
Tardieu F. 2003. Virtual plants: modelling as a tool for the genomics
of tolerance to water deficit. Trends in Plant Science 8, 9–14.
Tardieu F, Bruckler L, Lafolie F. 1992. Root clumping may affect
the root water potential and the resistance to soil–root water transport.
Plant and Soil 140, 291–301.
Tinker PB, Nye PH. 2000. Solute movement in the rhizosphere.
Oxford: Oxford University Press.
Tobin B, Cermak J, Chiatante D, et al. 2007. Towards
developmental modelling of tree root systems. Plant Biosystems 141,
481–501.
Tsegaye T, Mullins CE, Diggle AJ. 1995. An experimental
procedure for obtaining input parameters for the rootmap root
simulation program for peas (Pisum-sativum l). Plant and Soil 172,
1–16.
van Beek L, Wint J, Cammeraat L, Edwards J. 2005. Observation
and simulation of root reinforcement on abandoned mediterranean
slopes. Plant and Soil 278, 55–74.
van Noordwijk M, Brouwer G, Meijboom F, Do Rosario M,
Oliveira G, Bengough A. 2000. Trench profile techniques and core
break methods. In: Smit A, Bengough A, Engels C, Van Noordwijk M,
Pellerin S, Van de Geijn S, eds. Root methods: a handbook. Berlin:
Springer, 212–233.
van Noordwijk M, Spek L, de Willigen P. 1994. Proximal root
diameter as predictor of total root size for fractal branching models.
Plant and Soil 164, 107–117.
Waldron LJ. 1977. The shear resistance of root-permeated
homogeneous and stratified soil. American Journal of Soil Science
Society 41, 843–849.
Watanabe T, Yokozawa M, Emori S, Takata K, Sumida A,
Hara T. 2004. Developing a multilayered integrated numerical model
of surface physics–growing plants interaction (MINoSGI). Global
Change Biology 10, 963–982.
Weiner J, Damgaard C. 2006. Size-asymmetric competition and
size-asymmetric growth in a spatially explicit zone-of-influence model
of plant competition. Ecological Research 21, 707–712.
Wells CE, Eissenstat DM. 2001. Marked differences in
survivorship among apple roots of different diameters. Ecology 82,
882–892.
White PJ, Broadley MR, Greenwood DJ, Hammond JP. 2005.
Genetic modifications to improve phosphorus acquisition by roots.
Proceedings of the International Fertiliser Society 568. York, UK:
IFS.
Wilkinson S, Davies WJ. 2002. ABA-based chemical signalling: the
co-ordination of responses to stress in plants. Plant, Cell and
Environment 25, 195–210.
Wu L, McGechan MB, McRoberts N, Baddeley JA, Watson CA.
2007. SPACSYS: integration of a 3D root architecture component to
carbon, nitrogen and water cycling—model description. Ecological
Modelling 200, 343–359.
Zhang H, Forde BG. 1998. An arabidopsis MADS box gene that
controls nutrient-induced changes in root architecture. Science 279,
407–409.
Zienkiewicz OC, Taylor RL. 1998. The finite element method.
London: McGraw-Hill.
Continuous root growth models | 2143 at N
iedersaechsische Staats- u. U
niversitaetsbibliothek Goettingen on July 22, 2010
http://jxb.oxfordjournals.orgD
ownloaded from