f u n g a l b i o l o g y r e v i e w s 2 6 ( 2 0 1 2 ) 3 0 e3 8

journal homepage: www.elsevier.com/locate/fbr

Review

Modelling hyphal networks

Graeme P. BOSWELLa, Fordyce A. DAVIDSONb,*

a
Department of Computing and Mathematics, University of Glamorgan, Pontypridd, CF37 1DL, UK
b
Division of Mathematics, University of Dundee, Perth Road Dundee, DD1 4HN, UK

article info abstract

Article history: The indeterminate growth habit of fungal mycelial can produce massive organisms span-
Received 15 June 2011 ning kilometres, whereas the hypha, the modular building block of these structures, is
Accepted 3 February 2012 only a few microns in diameter. The qualitative and quantitative relationship between these
scales is difficult to establish using experimental methods alone and a large number of math-
Keywords: ematical models have been constructed to assist in the investigation of the multi-scale form
Anastomosis and function of filamentous fungi. Many such models operate at the colony-scale, represent-
Cellular automata ing the hyphal network as either a regular lattice or as a geometrically-unconstrained struc-
Lattice-free model ture that changes according to a minimal set of specified rules focussed on the fundamental
Mathematical model processes responsible for growth and function. In this review we discuss the historical devel-
Mycelium opment and recent applications of such models and suggest some future directions.
Translocation ª 2012 The British Mycological Society. Published by Elsevier Ltd. All rights reserved.

1. Introduction organisms (one clone of Armillaria gallica covers over 15 hect-

ares of forest (Smith et al., 1992)), whilst the modular building
Recent advances in genomics and the reactive development of block of these structures, the fungal hypha, is only a few
“systems biology” have driven many aspects of biological microns in diameter. To model the interaction of such
research in a direction heavily weighted towards computa- extremes of scale seems an almost overwhelming task. Not
tional, quantitative and predictive analysis. Mathematical too surprisingly then, most developments in modelling myce-
modelling is a key part in this development and it is unsur- lial fungi have been made by focussing on selected scales.
prising that it has played a significant role in expanding our At the macro-scale, the interaction of fungi with the envi-
understanding of the growth and function of the fungal myce- ronment forms the main focus. Variables in such models
lium. An excellent and extensive review of the applications of represent densities (or numbers) and the interaction of these
mathematical modelling to fungal growth, conducted up to densities is usually modelled via systems of ordinary or partial
the mid-1990s, can be found in Prosser (Prosser et al., 1995). differential equations. Examples include the modelling of
General modelling advances since that date are summarized carbon cycling in the environment (Lamour et al., 2000), fungal
in Davidson (2007) and Davidson et al. (2011). crop pathogens (Parnell et al., 2008) and biocontrol (Jeger et al.,
One of the main problems that faces modellers is the 2009; Cunniffe and Gilligan, 2009; Stevens and Rizzo, 2008).
choice of scale. In the study of fungal mycelia, the question At the other extreme of scale, much modelling work has
of scale is expressed in an extreme manner. The indetermi- focussed on hyphal tips where two main hypotheses have
nate growth habit of the mycelium can produce massive developed in parallel over recent decades. The steady-state

* Corresponding author.
E-mail addresses: gpboswel@glam.ac.uk (G. P. Boswell), fdavidson@maths.dundee.ac.uk (F. A. Davidson).
1749-4613/$ e see front matter ª 2012 The British Mycological Society. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.fbr.2012.02.002
Modelling hyphal networks 31

(SS) theory of Sietsma and Wessels, 1994 proposed that plastic (e.g. Regalado et al., 1996; Prosser and Trinci, 1979; Soddell
wall material is continually deposited at the hyphal apex and et al., 1994; Meskauskas et al., 2004a, b) and are often derived
cross-linked into a more rigid form over time. The vesicle from statistical properties of the experimental system under
supply centre (VSC) hypothesis (Bartnicki-Garcia et al., 1995; investigation. Some of these models (e.g. Prosser and Trinci,
Bartnicki-Garcia et al., 2001) predicts that the Spitzenko € rper, 1979) yield statistical properties close to those of real mycelia,
or equivalent structure, acts as a distribution point for vesicles whilst others (e.g. Meskauskas et al., 2004a) produce images
containing cell wall synthesizing materials. Turgor pressure is almost indistinguishable from real fungi and are therefore
also assumed to play some role in driving tip growth (Regalado very appealing. Significant advances can be made using these
et al., 1997; Bartnicki-Garcia and Oslewacz, 2002). A detailed types of model, for example in the testing of hypotheses con-
and extensive account of the development of the various theo- cerning basic growth architecture. In particular the models
ries regarding hyphal tip growth are given in Bartnicki-Garcia developed by Meskauskas et al. (which were developed into
and Oslewacz (2002), Goriely and Tabor (2003a, b). More a user-interactive experimental system) can consider
recently, Tindemans et al. (2006) modelled important details different species growing in 3-dimensional space and within
of the diffusive transfer of the vesicles from the Spitzenko € rper a variety of nutrient distributions. It must be noted, however,
to the hyphal wall and their subsequent fusion with the cell that in this modelling category there is the tendency to use
membrane. The maturation process resulting in the stiffening non-mechanistic rules to generate hyphal tip extension and
of the membrane is modelled by Eggen et al. (2011). Goriely and branching, i.e. the underlying mechanisms for growth are
Tabor (2008) provide an excellent overview of tip modelling to not modelled directly and are instead replaced by abstract
that date. branching and growth rules. Consequently, difficulties arise
The focus of this review is at the intermediate or “single in attempting to make and test hypotheses concerning
colony” scale where modelling approaches generally fall into changes in growth dynamics and mycelial function in
two categories. One strategy is to assume that the mycelium response to external factors. Moreover, because of this
is a continuum, the properties of which can be viewed in abstraction, it is difficult to choose parameter values in any
some sense as an average of the individual components a priori meaningful way. Furthermore, because of computa-
(much like in the modelling of fluid dynamics, for example). tional difficulties, it is only very recently that the two key
Such models have their roots in earlier work of Edelstein and processes of anastomosis and translocation, have been incor-
co-workers (see Edelstein, 1982; Edelstein and Segel, 1983; porated (see Boswell et al., 2007). These processes are crucial to
Edelstein-Keshet and Ermentrout, 1989). The models devel- mycelial development in general and in particular to growth
oped and analysed in e.g. Regalado et al. (1996) and Davidson in heterogeneous environments.
(1998), and references therein, and more recently by Lo  pez This class of discrete models can be further divided into
and Jensen (2002), Boswell et al. (2002); Boswell et al. (2003a, b) two important subgroups: lattice-based and lattice-free
and Falconer et al. (2006) all fall into this category. In these models. Essentially, the former assumes that the mycelial
studies, systems of equations (non-linear partial differential network is constrained to a predetermined grid or lattice.
equations) are derived that represent the (implicit or explicit) The latter allows the network to be free of a priori constraints,
interaction of fungal biomass and at least one growth- although this freely developing network has to be mapped
limiting substrate (e.g. a carbon source) as well as other factors onto a discrete grid for computational purposes at some point
(e.g. toxins). Such an approach is ideal when modelling dense in the calculations. Both approaches have advantages and
mycelia, for example growth in Petri dishes or on the surfaces disadvantages. Clearly the former restricts the topology of
of solid substrates such as foodstuffs, plant surfaces and the network, whilst the latter can be computationally expen-
building materials. This modelling strategy has, for example, sive. We now outline the structure and development of these
allowed the study of biomass distribution within the mycelium two approaches in turn.
in homogeneous and heterogeneous conditions, translocation
in a variety of habitat configurations as well as certain func- Lattice-based models
tional consequences of fungal growth including acid produc-
tion. The models developed by Boswell et al. (2002); Boswell A well-established and computationally efficient approach to
et al. (2003a, b), Boswell (2008) are the distillation of much of modelling filamentous fungi has been to use a regular lattice
the modelling work conducted over the previous 10 years. as the basis for the mycelial network. Such models are essen-
A second category of colony-scale model is based on a discrete tially forms of cellular automata (CA) and are typically discrete
modelling approach, in which individual hyphae are identi- in time, space and state. By formulating a series of carefully
fied. It is this second approach, which explicitly details the selected stochastic rules applied at the local level, the status
development of the hyphal network, that will form the focus of each element in the lattice changes over regular time steps
of the remainder of our review. representing the growth and development of the fungal
network.
An early CA model was devised by Ermentrout and
2. Modelling of hyphal networks Edelstein-Keshet (1993) in which a regular grid comprising
square cells was used to represent the growth environment.
Historical context Despite exceptionally simple growth rules covering only the
processes of tip growth, branching and anastomosis, a variety
Discrete models for the development of hyphal networks of networks were created that captured the fractal-like struc-
generally take the form of computer-generated simulations ture of filamentous fungi implicating, at least in uniform
32 G. P. Boswell, F. A. Davidson

growth conditions, the fundamental processes of tip forma- upon the network. Nutrient translocation in fungal mycelia
tion and movement in mycelial growth (Fig. 1). Indeed, the was used as a good example. While the resultant network
model derived by Ermentrout and Edelstein-Keshet (1993) structure did not represent a mycelium, the concentration of
provided the template for numerous other lattice-based the internalized nutrients was shown to have the same distri-
models that describing the growth, and later function, of butions as radioactively-labelled amino acid within Phanero-
fungal mycelia. chaete velutina.
An entirely different approach was adopted by Lo  pez and A closely related approach had been previously developed
Jensen (2002). Building on a previous model used to investigate in Boswell et al. (2007). This hybrid mathematical model repre-
the changing morphology of colony peripheries (Lo  pez and sented a growth-promoting substrate (a carbon source) as
Jensen, 1998) Lo pez and Jensen constructed a stochastic model a continuous variable and the biomass network as a discrete
in which a hypothetical inhibiting waste material produced by structure. The explicit inclusion of internal and external
the fungus itself diffused and affected the ability of the fungus substrate allowed for the modelling of the fundamental rela-
to grow elsewhere. While no explicit hyphal network was tionship between nutrient availability and hyphal tip growth.
simulated in their model, predictions were made on the This enabled, for the first time, simulation of the growth of
morphology of the growth fronts of simulated mycelia in hyphal networks in response to heterogeneous environments.
response to different nutrient concentrations (that were Moreover, this model was able to predict a functional conse-
assumed uniform and constant through the simulation) and quence of such growth, namely the acidification of the
different sensitivities to the inhibitory substance. These surrounding environment. Finally, another advantage of this
predictions were consistent with experimental observations method was that it used parameter values and rules for
reported in Matsuura and Miyazima (1992); Matsuura and growth and metabolism drawn directly from the calibrated
Miyazima (1993). and tested continuum model developed by Boswell et al.
Smith et al. (2011) adopted a novel approach by construct- (2003b).
ing a network automata (NA) in which connections in network The simulated network developed according to stochastic
(rather than states of “cells” in a CA) were updated at each rules calibrated for the fungus Rhizoctonia solani in which tip
time step. Such an approach was shown to be particularly extension and branching were related to internally located
suited to the situation where the creation of connections in substrate, which itself was translocated, by both passive diffu-
the network was coupled to a dynamic process occurring sion and active metabolically-driven processes. As a model tip
moved between neighbouring nodes on a triangular lattice, the
edge connecting these nodes became an active hypha involved
in the uptake of substrate and its subsequent translocation
through the network, much like the NA framework later
proposed by Smith et al. (2011). The model was applied to
a series of planar environments representing uniform condi-
tions, nutritionally heterogeneous conditions and soil slices
that exhibited both nutritional and structural heterogeneities.
It was shown that substrate translocation and the physical
environment significantly influenced the structure of the
networks produced and the extent of zones of acidification
that preceded the leading edge of the biomass network (Fig. 2).
A three-dimensional version of this model was subse-
quently developed by Boswell (2008) to simulate the formation
of mycelia in soil-like systems using a face-centred cubic
lattice. A related approach used to model the growth of aerial
hyphae has been recently adopted by Coradin et al. (2011) to
further understand the role of filamentous fungi in solid state
fermentation processes (Nopharatana et al., 1998).
Irrespective of the processes represented in lattice-based
models (and indeed the lattice structure), the regular geom-
etry imposes certain limitations on the network constructed.
Computationally, a regular geometry has numerous advan-
tages (compared to an irregular geometry) principally because
there are a finite number of orientations adopted by lengths of
biomass and hence a finite number of rules governing the
development of the biomass structures. While the use of
regular geometries may be suitable for certain applications,
in other instances, e.g. the representation of mycelial growth
in response to various tropisms (Gooday and Carlile, 1975;
Fig. 1 e Output from the lattice-based mycelial network Fomina et al., 2000), such a regular geometry may fail to suffi-
model of Ermentrout and Edelstein-Keshet (1993). Image ciently capture the complex behaviour exhibited. Therefore,
reproduced with permission. models that represent mycelial networks using non-regular
Modelling hyphal networks 33

Fig. 2 e Simulations from the calibrated model of Boswell et al., (2007). A two-dimensional representation of non-saturated
soils where a water film surrounds soil particles (denoted by black cells) with air-filled gaps elsewhere (open cells). The
network is represented by solid lines. The predicted pH of the environment is shown over regular time intervals (dark red
indicates regions with low pH).

geometries (i.e. are lattice-independent) have been developed In a pain of papers Yang et al. (1992a, b) developed a hybrid
alongside lattice-based models. model that generated an explicit network representing the
growth of both filamentous fungi and mycelial bacteria. Their
Lattice-free models model had a similar form to that developed by Cohen, 1967
except it had a mechanistic underpinning. Principally, the
Although strictly not an mathematical model for the develop- rules governing branching and hyphal tip extension were
ment of fungal mycelia, the first attempt to develop a lattice- dependent on an internally located and self-produced mate-
free model for networks was conducted by Cohen (1967). He rial that was transported by diffusion through the developing
developed a generic model of branching networks in two network.
spatial dimensions and considered filamentous fungi as one In a series of papers Stack et al. (1987); Knudsen et al. (1991,
such application. While this computer simulation produced 2006) developed and calibrated an individual-based computer
images highly reminiscent of certain fungal mycelia, it did simulation that described the three-dimensional growth of
not relate the rules of the network’s development to available a biological control fungus in soils. Their models were again
nutrients and therefore its predictive ability was limited. based on that of Cohen (1967) where the fungal mycelium
None-the-less, despite these important limitations, this pio- was represented by a series of connected straight line
neering approach created a template upon which a series of segments where each segment contained information about
improved and revised lattice-free mathematical models its position, orientation, connectivity and relative nutrient
were developed. status.
34 G. P. Boswell, F. A. Davidson

Fig. 3 e Examples of networks produced using the neighbour-sensing model of Meskauskas et al. (2004b). (Images obtained
from the online simulation at http://www.world-of-fungi.org/Models/mycelia_3D/index.html.)

A related approach to modelling the development in represented by a collection of line segments but unlike in
three dimensions was adopted by Meskauskas et al. (2004a, previous models, a variety of tropisms were considered
b) that described a neighbour-sensing mathematical model including negative autotropism (measured with respect to
for hyphal growth (Fig. 3). The fungal network was again the density of other line segments), galvanotropism (based

Fig. 4 e Biomass networks developing from a continually-replenished nutrient source. Simulations produced using the
model of Carver and Boswell (2008). Figures (a)e(d) correspond to networks generated where the only difference is an
increase in the diffusive translocation component.
Modelling hyphal networks 35

on self-generated electric fields that either align or diverge were modelled. Previously, model tips were typically assumed
adjacent hyphae) and gravitropism (with growth either to change direction according to a random variable drawn
following or opposing the gravitational field). from a normal distribution and, therefore, statistical proper-
The first lattice-free model to incorporate the key property ties obtained from experimental data were used to calibrate
of anastomosis was developed by Carver and Boswell (2008). the reorientation process, meaning the predictive aspects of
Their approach essentially combined the planar model of the models were limited to situations resembling the calibra-
Cohen (1967) with the nutrient reallocation techniques of tion experiment. To overcome this limitation, Hopkins and
Yang et al. (1992b) and checked for the crossing of line Boswell (2012) utilized a (biased) circular random walk to
segments at each stage of the simulation. During a regular model tip orientation and related this to the corresponding
time step, each tip could extend a fixed distance with a proba- FokkerePlanck partial differential equation (which describes
bility that increased with the substrate concentration in the statistical properties of the random walk). When simulated
corresponding line segment (similar to Boswell et al., 2007) in heterogeneous conditions, the model predicted the emer-
and in a direction normally-distributed from its previous gence of pathways in the biomass network radiating out
orientation. The simulated interconnected networks closely from substrate sources that contained significantly more
resembled genuine mycelia and the morphological effects of internal substrate than other model hyphae in the structure
increasing the rate of nutrient translocation resulted in (Fig. 5). Such a feature is consistent with the emergence of
increasingly dense biomass structures at the colony periphery fungal cords (Boddy, 1993, 1999) and noticeably did not require
(Fig. 4). While this model was only partially calibrated any “global” input. Instead, these pathways arose through an
for R. solani and applied to a highly specific experimental initially stochastic but then self-reinforcing locally-applied
system corresponding to outgrowth from a nutrient source, process, a concept independently suggested in a recent
it provided the basis for further mathematical models capable modelling investigation by Heaton et al. (2010) who used
of simulating fungal growth and function in more complex scanned images of fungal colonies and related the internal
environments. flow of material to that of currents in electric circuits.
Boswell and Hopkins (2008) and Hopkins and Boswell (2012) A variety of lattice-free approaches have been developed
extended the work of Carver and Boswell (2008) to allow for over the last couple of decades, each offering differing degrees
the simulation of planar growth in arbitrary nutritional condi- of flexibility and applicability. Almost all such models have
tions. Similar to the lattice-based model of Boswell et al. (2007), relied on a mixture of deterministic and stochastic elements.
a generic substrate was assumed to exist in two forms: (i) free The modelling complexity and computation costs associated
in the external environment where it could diffuse and (ii) with such models have often forced compromises and
contained within the biomass network where it was translo- possibly important biological features have been simplified
cated and used to fuel tip extension. An important advance or even omitted. The challenge for modellers is to find a mean-
made by Hopkins and Boswell (2012) over earlier lattice-free ingful balance between biological function and mathematical
models was the manner in which the orientation of hyphae computability.

3. Current approaches and applications

It is well-established that certain fungi can be used as agents in

bioremediation (Sayer et al., 1995; Gadd, 2001), but recent inves-
tigations have shown they can also be used as conduits to
significantly speed up the dispersal of other micro-organisms,
capable of biorestoration, through soil systems (Kohlmeier
et al., 2005; Banitz et al., 2011). In order to make best use of fungi
in such settings, it is essential to first predict the growth
dynamics of hyphal networks and then study how these can
be advantageously manipulated. To this end, the model in
Hopkins and Boswell (2012) has recently been adapted (Hopkins
and Boswell, unpublished) in the following manner.
The model is being applied to consider the problem of
a pollutant diffusing in an essentially planar environment
(e.g. shallow soil) (Fig. 6). Consistent with experimental obser-
vations and previously successful modelling approaches,
model tips are assumed to reorientate themselves away
Fig. 5 e Mycelial growth from a nutrient source into from both existing biomass (Hickey et al., 2002) and, indepen-
a nutrient-replete environment. Simulation using the model dently, the pollutant (Fomina et al., 2000), with the latter at
of Hopkins and Boswell (2012). The thick line segments in a rate dependent on the pollutant concentration and level of
the biomass network contain greater concentrations of an internal substrate, si. Thus, the biomass resilience to the
internal substrate than the thin line segments suggesting effects of the pollutant is increased with nutrient availability
the emergence of fungal cords from the nutrient source. (Gadd et al., 2001; Fomina et al., 2003).
36 G. P. Boswell, F. A. Davidson

Fig. 6 e Network structure after 2 d growth in the presence of a pollutant under three different configurations of an external
substrate. Simulations produced using the model of Hopkins and Boswell (2012). The initial pollutant concentration is
confined to the circular region and the locations of the external substrate are represented by the square blocks that are either
located (a) at the site of inoculation, (b) at the inoculation site and at edge of the diffusing pollutant, and (c) at the inoculation
site and at the pollutant source. The mean position of the edge of the network (dotted line) and the position of where the
network density first exceeded unity (solid line, representing the “functional” part of the network (Hopkins and Boswell,
2012)) were calculated from three replicates over the duration of the simulation along a rectangular strip between the
inoculation site and pollutant source where (d)e(f) correspond to the configurations in (a)e(c) respectively.

At discrete time steps, each model tip is assumed to reor- 2012; Plank and Sleeman 2004, for full details). Once the
ientate itself according to a preferred direction of growth new orientation is determined, provided there is sufficient
depending on the pollutant and biomass densities. Local internal substrate, the model tip advances a fixed distance
gradients dictate the direction of least biomass and and if there is a collision with an existing line segment it
pollutant, generating a preferential direction of growth qc undergoes fusion (representing anastomosis in the network).
and qp respectively (see Hopkins and Boswell, 2012, for A new line segment is then generated between the old and
details). The bias of reorientation of a model tip of alignment updated position of the model tip. The production of new
q is given by model tips, corresponding to subapical branching, is included

and since turgor pressure has been implicated in branching
mðqÞ ¼ dp ðsi Þ sin q  qp  dc sinðq  qc Þ
(Gow and Gadd, 1995; Riquelme and Bartnicki-Garcia, 2004)
where dp ðsi Þ ¼ g1 pð1  si =ðg2 þ si ÞÞ and dc are the relative reor- it is assumed the branching rate is zero if si is less than a crit-
ientation rates in response to differences in the current and ical value and otherwise increases proportionally with si.
preferred growth alignment and g1, g2, dc are constants. (If Internal substrate is updated, representing growth costs, by
qp ¼ qc then this is the preferred direction of growth. If the uptake of new resources and by translocation, which
qc s qp the model tip essentially seeks a compromise in the comprises both diffusive directed components (towards the
growth orientation between the different tropisms. In both nearest tip) (see Hopkins and Boswell, 2012, for details on
instances, stochastic variations about the preferred direction the implementation of this process).
are also included.) The reorientation process itself is simu- The growth domain represents a polluted environment
lated by a biased random walk obtained from a FokkerePlanck prior to the introduction of fungal biomass. To investigate
equation that relates statistical properties of the tip reorien- the effects of the augmentation of nutrients, three configura-
tation process to external stimuli (see Hopkins and Boswell, tions have been considered:
Modelling hyphal networks 37

1. single block at “inoculation” site (Fig. 6a), references

2. two blocks, first at “inoculation” site and second between
“inoculation” site and pollutant source (Fig. 6b),
3. two blocks, first at “inoculation” site and second at the Banitz, T., Fetzer, I., Johst, K., Wick, L.Y., Harms, H., Frank, K.,
pollutant source (Fig. 6c). 2011. Assessing biodegradation benefits from dispersal
networks. Ecol. Model 222, 2552e2560.
The total amount of augmented external substrate is taken Bartnicki-Garcia, S., 2002. Hyphal tip growth, outstanding ques-
tions. In: Oslewacz, H.D. (Ed.), Molecular Biology of Fungal
to be equal in all instances (its concentration is halved when
Development, pp. 29e58.
divided between two substrate blocks). The model was simu- Bartnicki-Garcia, S., Bartnicki, D.D., Gierz, G., Lopez-Franco, R.,
lated with three replicates for each configuration over a period Bracker, C.E., 1995. Evidence that Spitzenko € rper behavior
of time representative of 2 d and the biomass density along determines the shape of a fungal hypha: a test of the hyphoid.
a rectangular strip between the site of inoculation and the Exp. Mycol. 19, 153e159.
pollutant centre was calculated. Bartnicki-Garcia, S., Bartnicki, D.D., Gierz, G., Lopez-Franco, R.,
Bracker, C.E., 2001. A three-dimensional model of fungal
In all configurations, the biomass initially expanded
morphogenesis based on the vesicle supply center concept.
outward in a radially-symmetric manner. After this transient
J. Theor. Biol. 208, 151e164.
feature, line segments extending away from the pollutant Boddy, L., 1993. Saprotrophic cord-forming fungi: warfare strate-
continued to extend and branch but those line segments gies and other ecological aspects. Mycol. Res. 97, 641e655.
initially extending towards the pollutant source typically expe- Boddy, L., 1999. Saprotrophic cord-forming fungi: meeting the
rienced a sharp reorientation due to the presence of the challenge of heterogeneous environments. Mycologia 91, 13e32.
diffusing pollutant (Fig. 6a, c). However, when the external Boswell, G.P., Hopkins, S., 2008. Linking hyphal growth to colony
dynamics: spatially explicit models of mycelia. Fungal Ecol 1,
substrate was divided evenly between the site of inoculation
143e154.
and midway between that and the centre of the pollutant Boswell, G.P., Jacobs, H., Davidson, F.A., Gadd, G.M., Ritz, K., 2002.
concentration, the resultant increased level of internal Functional consequences of nutrient translocation in mycelial
substrate within the biomass, caused increased branching fungi. J. Theor. Biol. 217, 459e477.
and tips resilience to the pollutant (e.g. Fig. 6b). By locating Boswell, G.P., Jacobs, H., Davidson, F.A., Gadd, G.M., Ritz, K.,
where the leading edge of the biomass density first exceeded 2003a. A positive numerical scheme for a mixed-type partial
a critical value (representing the extent of the “functionally” differential equation model for fungal growth. Appl. Math.
Comput. 138, 321e340.
capable mycelium (Hopkins and Boswell, 2012)) it was observed
Boswell, G.P., Jacobs, H., Davidson, F.A., Gadd, G.M., Ritz, K.,
that the simulated network extended closer to the pollutant 2003b. Growth and function of fungal mycelia in heteroge-
source when external substrate was available at the pollutant neous environments. Bull. Math. Biol. 65, 447e477.
edge than when it was distributed elsewhere (Fig. 6d, e, f). Boswell, G.P., Jacobs, H., Ritz, K., Gadd, G.M., Davidson, F.A., 2007.
These simulations suggest that the extent of mycelial The development of fungal networks in complex environ-
growth can be advantageously manipulated when subjected ments. Bull. Math. Biol. 69, 605e634.
Boswell, G.P., 2008. Modelling mycelial networks in structured
to toxic conditions. Crucially, the model predicts that a simple,
environments. Mycol. Res. 112, 1015e1025.
uniform addition of extra nutrients would not be sufficient to
Carver, I., Boswell, G.P., 2008. A lattice-free model of translocation
promote a change in mycelial extent; instead the location of induced outgrowth in mycelial fungi. IAENG Int. J. Appl. Math.
the extra nutrients relative to the pollutant is fundamentally 38, 173e179.
important. These and related predictions may assist further Cohen, D., 1967. Computer simulation of biological pattern
refinements in the application of fungi in bioremediation. generation processes. Nature 216, 246e248.
Coradin, J.H., Braun, A., Viccini, G., de Lima Luz Jr., L.F.,
Krieger, N., Mitchell, D.A., 2011. A three-dimensional discrete
lattice-based system for modeling the growth of aerial hyphae
4. Future challenges of filamentous fungi on solid surfaces: a tool for investigating
micro-scale phenomena in solid-state fermentation. Biochem.
The delivery of wall building materials to the tip, movement of Eng. J. 54, 164e171.
organelles over large distances at high speed and nuclear mix- Cunniffe, N., Gilligan, C.A., 2009. Scaling from mycelial growth to
ing are all facets of mature networks that invite further inves- infection dynamics: a reaction diffusion approach. Fungal Ecol
tigation. At the other end of the development cycle, early germ 1, 133e142.
Davidson, F.A., Boswell, G.P., Fisher, M.W.F., Heaton, L.,
tube growth and anastomosis is only now beginning to be
Hofstadler, D., Roper, M., 2011. Mathematical modelling of
understood (Roca et al., 2010). We suggest that a major chal- fungal growth and function. IMA Fungus 2, 33e37.
lenge for fungal biology is to link this increasing body of infor- Davidson, F.A., 1998. Modelling the qualitative response of fungal
mation at the micro-scale to the large scale form and function mycelia to heterogeneous environments. J. Theor. Biol. 195,
of hyphal networks. 281e292.
The increasingly quantitative nature of experimental data Davidson, F.A., 2007. Mathematical modelling of mycelia: a ques-
tion of scale. Fungal Biol. Rev. 21, 30e41.
regarding tip growth, cytoplasmic flow, nutrient transport
Edelstein, L., Segel, L.A., 1983. Growth and metabolism in mycelial
and the imaging of network architecture will undoubtedly
fungi. J. Theor. Biol. 104, 187e210.
underpin the development of new models that are genuinely Edelstein, L., 1982. The propagation of fungal colonies: a model
predictive. The authors predict that mathematical models of for tissue growth. J. Theor. Biol. 98, 679e701.
the type discussed here, where the dynamic formation and Edelstein-Keshet, L., Ermentrout, B., 1989. Models for branching
function of the mycelium is explicitly captured, could play networks in two dimensions. SIAM J. Appl. Math. 49,
a significant role in linking information across scales. 1136e1157.
38 G. P. Boswell, F. A. Davidson

Eggen, E., de Keijzer, N., Mulder, B., 2011. Self-regulation in tip- Meskauskas, A., McNulty, L.J., Moore, D., 2004a. Concerted regu-
growth: the role of cell wall ageing. J. Theor. Biol. 283, 113e121. lation of all hyphal tips generates fungal fruit body structures:
Ermentrout, G.B., Edelstein-Keshet, L., 1993. Cellular automata experiments with computer visualizations produced by a new
approaches to biological modelling. J. Theor. Biol. 160, 97e133. mathematical model of hyphal growth. Mycol. Res. 108,
Falconer, R.E., Brown, J.L., White, N.A., Crawford, J.W., 2006. 341e353.
Biomass recycling and the origin of phenotype in fungal Meskauskas, A., Fricker, M.D., Moore, D., 2004b. Simulating colo-
mycelia. Proc. Roy. Soc. B 272, 1727e1734. nial growth of fungi with the neighbour-sensing model of
Fomina, M., Ritz, K., Gadd, G.M., 2000. Negative fungal chemot- hyphal growth. Mycol. Res. 108, 1241e1256.
ropism to toxic metals. FEMS Microbiol. Lett. 193, 207e211. Nopharatana, M., Howes, T., Mitchell, D.A., 1998. Modelling
Fomina, M., Ritz, K., Gadd, G.M., 2003. Nutritional influence on the fungal growth on surfaces. Biotechnol. Tech 12, 313e318.
ability of fungal mycelia to penetrate toxic metal-containing Parnell, S., Gilligan, C.A., Lucas, J.A., Bock, C.H., van den Bosh, F.,
domains. Mycol. Res. 107, 861e871. 2008. Changes in fungicide sensitivity and relative species
Gadd, G.M., Ramsay, L., Crawford, J.W., Ritz, K., 2001. Nutri- abundance: Oculimacula yallundae and O. acuformis populations
tional influence on fungal colony growth and biomass (eyespot disease of cereals) in western Europe. Plant Pathology
distribution in response to toxic metals. FEMS Microbiol. 57, 509e517.
Let 204, 311e316. Plank, M.J., Sleeman, B., 2004. Lattice and non-lattice models for
Gadd, G.M. (Ed.), 2001. Fungi in Bioremediation. Cambridge tumour angiogenesis. Bull. Math. Biol. 66, 1785e1819.
University Press, Cambridge, ISBN 0-521-78119-1, p. 481. Prosser, J.I., Trinci, A.P.J., 1979. A model for hyphal growth and
Gooday, G.W., 1975. Chemotaxis and chemotrophism in fungi and branching. J. Gen. Microbiol. 111, 153e164.
algae. In: Carlile, M.J. (Ed.), Primitive Sensory and Communi- Prosser, J.I., 1995. Mathematical modelling of fungal growth. In:
cation Systems. Academic Press, London, pp. 155e204. Gow, N.A.R., Gadd, G.M. (Eds), The Growing Fungus. Chapman
Goriely, A., Tabor, M., 2003a. Self-similar tip growth in filamen- and Hall, London, pp. 319e335.
tary organisms. Phys. Rev. Lett. 90, 108101-1e108101-4. Regalado, C.M., Crawford, J.W., Ritz, K., Sleeman, B.D., 1996. The
Goriely, A., Tabor, M., 2003b. Biomechanical models of hyphal origins of spatial heterogeneity in vegetative mycelia: a reac-
growth in actinomycetes. J. Theor. Biol. 222, 211e218. tion-diffusion model. Mycol. Res. 100, 1473e1480.
Goriely, A., Tabor, M., 2008. Mathematical modeling of hyphal tip Regalado, C.M., Sleeman, B.D., Ritz, K., 1997. Aggregation and collapse
growth. Fungal Biol. Rev. 22, 77e84. of fungal wall vesicles in hyphal tips: a model for the origin of the
Gow, N., Gadd, G.M. (Eds), 1995. The Growing Fungus. Chapman Spitzenko € rper. Phil. Trans. Roy. Soc. B 352, 1963e1974.
and Hall, London. Riquelme, M., Bartnicki-Garcia, S., 2004. Key differences between
Heaton, L.L.M., Lopez, E., Maini, P.K., Fricker, M.D., Jones, N.S., lateral and apical branching in hyphae of Neurospora crassa.
2010. Growth-induced mass flows in fungal networks. Proc. Fungal Genet. Biol. 41, 842e851.
Roy. Soc. B 277, 3265e3274. Roca, M.G., Kuo, H.-C., Lichius, A., Freitag, M., Read, N.D., 2010.
Hickey, P.C., Jacobson, D., Read, N.D., Glass, N.L., 2002. Live-cell Nuclear dynamics, mitosis and the cytoskeleton during the
imaging of vegetative hyphal fusion in Neurospora crassa. early stages of colony initiation in Neurospora crassa. Eukaryot
Fungal Genet. Biol. 37, 109e119. Cell 9, 1171e1183.
Hopkins, S., Boswell, G.P., 2012. Mycelial response to spatiotemporal Sayer, J.A., Raggett, S.L., Gadd, G.M., 1995. Solubilization of
nutrient heterogeneity: a velocity-jump mathematical model. insoluble metal compounds by soil fungi: development of
Fungal Ecol 5, 124e136. a screening method for solubilizing ability and metal toler-
Jeger, J.M., Jeffries, P., Elad, Y., Xu, X.M., 2009. A generic theoret- ance. Mycol. Res. 99, 987e993.
ical model for biological control of foliar plant diseases. Sietsma, J.H., Wessels, J.G.H., 1994. Apical wall biogenesis. In:
J. Theor. Biol. 256, 201e214. Mycota. Growth, Differentiation and Sexuality, vol. 1.
Knudsen, G.R., Stack, J.P., 1991. Modeling growth and dispersal of Smith, M.L., Bruhn, J.N., Anderson, J.B., 1992. The fungus Armil-
fungi in natural environments. In: Arora, D.K., Rai, B., laria bulbosa is among the largest and oldest living organisms.
Mukerji, K.G., Knudsen, G.R. (Eds), Handbook of Applied Nature 356, 428e431.
Mycology. Soil and Plants, vol. I. Marcel Dekker, Inc., New Smith, D.M.D., Onnela, J.-P., Lee, C.F., Fricker, M.D., Johnson, N.F.,
York, pp. 625e645. 2011. Network automata: coupling structure and function in
Knudsen, G.R., Stack, J.P., Schuhmann, S.O., Orr, K., LaPaglia, C., dynamic networks. Adv. Complex Syst 14, 317e339.
2006. Individual-based approach to modeling hyphal growth Soddell, F., Seviour, R., Soddell, J., 1994. Using Lindenmayer
of a biocontrol fungus in soil. Phytopathology 96, 1108e1115. systems to investigate how filamentous fungi may produce
Kohlmeier, S., Smits, T.H., Ford, R.M., Keel, C., Harms, H., round colonies. Complex. Int. 2. available online: http:
Wick, L.Y., 2005. Taking the fungal highway: mobilization of //www.csu.edu.au/ci/vol2/f_soddel/f_soddel.html.
pollutant-degrading bacteria by fungi. Environ. Sci. Technol. Stack, J.P., Knudsen, G.R., Koch, D.O., 1987. A computer simula-
39, 4640e4646. tion model to predict the dispersal of biocontrol fungi in soil.
Lamour, A., van den Bosch, F., Termorshuizen, A.J., Jeger, M.J., Phytopathology 77, 1771.
2000. Modelling the growth of soil-borne fungi in response to Stevens, L., Rizzo, D.M., 2008. Local adaptation to biocontrol
carbon and nitrogen. IMA J. Math. Appl. Med. Biol. 17, 329e346. agents: A multi-objective data-driven optimization model for
Lo pez, J.M., Jensen, H.J., 1998. Nonequilibrium roughening tran- the evolution of resistance. Ecol. Complex 5, 252e259.
sition in a simple model of fungal growth in 1þ1 dimensions. Tindemans, S.H., Kern, N., Mulder, B.M., 2006. The diffusive
Phys. Rev. Lett. 81, 1734e1737. vesicle supply center model for tip growth in fungal hyphae.
Lo pez, J.M., Jensen, H.J., 2002. Generic model of morphological J. Theor. Biol. 238, 937e948.
changes in growing colonies of fungi. Phys. Rev. E 65 (2), 021903. Yang, H., Reichl, U., King, R., Gilles, E.D., 1992. Measurement and
Matsuura, S., Miyazima, S., 1992. Self-affine fractal growth front simulation of the morphological development of filamentous
of Aspergillus oryzae. Physica A 191, 30e34. microorganisms. Biotechnol. Bioeng 39, 44e48.
Matsuura, S., Miyazima, S., 1993. Colony of the fungus Aspergillus Yang, H., King, R., Reichl, U., Gilles, E.D., 1992. Mathematical
oryzae and self-affine fractal geometry of growth fronts. Frac- model for apical growth, septation and branching of mycelial
tals 1, 11e19. microorganisms. Biotechnol. Bioeng 39, 49e58.