Hyperbolic and Kinetic Models For Self Organised Biological Aggregations A Modelling and Pattern Formation Approach Raluca Eftimie
Hyperbolic and Kinetic Models For Self Organised Biological Aggregations A Modelling and Pattern Formation Approach Raluca Eftimie
Hyperbolic and Kinetic Models For Self Organised Biological Aggregations A Modelling and Pattern Formation Approach Raluca Eftimie
https://textbookfull.com/product/python-debugging-for-ai-machine-
learning-and-cloud-computing-a-pattern-oriented-approach-1st-
edition-vostokov/
https://textbookfull.com/product/from-mafia-to-organised-crime-a-
comparative-analysis-of-policing-models-1st-edition-dr-anna-
sergi-auth/
https://textbookfull.com/product/formation-of-knbo3-thin-films-
for-self-powered-reram-devices-and-artificial-synapses-tae-ho-
lee/
https://textbookfull.com/product/feature-engineering-and-
selection-a-practical-approach-for-predictive-models-1st-edition-
max-kuhn/
Combined Discrete and Continual Approaches in
Biological Modelling Alexander E. Filippov
https://textbookfull.com/product/combined-discrete-and-continual-
approaches-in-biological-modelling-alexander-e-filippov/
https://textbookfull.com/product/computed-tomography-of-the-lung-
a-pattern-approach-2nd-edition-johny-a-verschakelen/
https://textbookfull.com/product/radiology-illustrated-chest-
radiology-pattern-approach-for-lung-imaging-2nd-edition-lee/
https://textbookfull.com/product/flow-modelling-and-control-in-
pipeline-systems-a-formal-systematic-approach-sina-razvarz/
https://textbookfull.com/product/flow-modelling-and-control-in-
pipeline-systems-a-formal-systematic-approach-sina-razvarz-2/
Lecture Notes in Mathematics 2232
Mathematical Biosciences Subseries
Raluca Eftimie
Hyperbolic and
Kinetic Models
for Self-organised
Biological
Aggregations
A Modelling and Pattern Formation
Approach
Lecture Notes in Mathematics 2232
Editors-in-Chief:
Jean-Michel Morel, Cachan
Bernard Teissier, Paris
Advisory Board:
Michel Brion, Grenoble
Camillo De Lellis, Princeton
Alessio Figalli, Zurich
Davar Khoshnevisan, Salt Lake City
Ioannis Kontoyiannis, Athens
Gábor Lugosi, Barcelona
Mark Podolskij, Aarhus
Sylvia Serfaty, New York
Anna Wienhard, Heidelberg
In recent years, the role of mathematics in the life sciences has evolved a long
way from the role it played in the 1970’s, in the early days of “biomathematics”, and
is a somewhat different one now, and its perception by the mathematical community
is also different. We feel it is important for the Lecture Notes in Mathematics to
reflect this and thus underline the immense significance of the life sciences as a
field of application and interaction for mathematics in the 21st century.
We are particularly interested in going far beyond the traditional areas in which
mathematics was applied to ecology, such as population dynamics, and would like
to attract publications in areas such as cell growth, protein structures, physiology,
vision, shape recognition & gestalt theory, neural dynamics, genomics, perhaps also
some statistical aspects (this list is non-exhaustive).
123
Raluca Eftimie
Division of Mathematics
University of Dundee
Dundee, UK
Mathematics Subject Classification (2010): 92-01, 92-02, 92C15, 92D50, 35C07, 35Lxx, 35Q20, 35Q92,
35R09, 35R60, 37G40, 58J55, 65Nxx
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
To Georgi
Preface
vii
viii Preface
numerical investigation of these mathematical models. Prof. Mark Lewis was very
supportive to extend a review article on hyperbolic and kinetic models for self-
organised biological aggregations (published in 2012 in the Journal of Mathematical
Biology) into a book. I am also very grateful to Prof. Pietro-Luciano Buono who
introduced me to the fascinating field of equivariant bifurcation theory: without
our collaboration on classifying the various spatial and spatio-temporal patterns
exhibited by 1D nonlocal hyperbolic systems, this monograph would not have
come to light. Prof. Thomas Hillen opened my eyes to the theory of hyperbolic
conservation laws (through the postgraduate courses he taught at the University
of Alberta while I was a PhD student). I must also thank Prof. Jose Carrillo de
la Plata and Prof. Nicola Bellomo who introduced me to their research on higher
dimensional kinetic equations (which led to our current collaborations). Many more
colleagues and collaborators, among which I mention Prof. Razvan Fetecau and
Prof. Kees Weijer, have influenced over the past years my research on pattern
formation in biological systems, which was the starting point of this monograph.
Finally, I would like to thank the editorial staff at Springer, in particular Dr. Eva
Hiripi, for their approachability and help with this book.
References
1. P. the Elder, Birds which take their departure from us, and whither they go; The
thrush, the blackbird, and the starling – Birds which lose their feathers during
their retirement – The turtle-dove and the ring-dove – The flight of starlings and
swallows, in The Natural History (Taylor and Francis, London, 1855). Translated
by John Bostock, M.D., F.R.S., and H.T. Riley, Esq. B.A.
2. C.M. Breder, Ecology 35(3), 361 (1954)
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
1.1 Modelling Self-organised Aggregation and Movement .. . . . . . . . . . . . . 1
1.1.1 Individual-Based Models . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 4
1.1.2 Kinetic and Macroscopic Models.. . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
1.2 The Importance of Communication for Self-organised
Biological Behaviours .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 20
1.3 Overview .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 23
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 32
2 A Short Introduction to One-Dimensional Conservation Laws. . . . . . . . . 37
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 37
2.2 Fundamental Results for Systems of Conservation Laws.. . . . . . . . . . . 40
2.2.1 Travelling Waves, Rarefaction Waves, Shocks
and Contact Discontinuities .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 45
2.2.2 The Rankine-Hugoniot Jump Condition .. . . . . . . . . . . . . . . . . . . . 47
2.2.3 Admissibility Conditions: Entropy, Vanishing
Viscosity and Speed Stability . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 48
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 52
3 One-Equation Local Hyperbolic Models . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 55
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 55
3.2 First-Order Traffic Models . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 57
3.3 Second-Order Traffic Models . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 62
3.4 Third-Order Traffic Models . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 68
3.5 Traffic Models that Include Reaction Terms . . . . .. . . . . . . . . . . . . . . . . . . . 69
3.6 Advection-Reaction Equations for Animal Population
Dynamics.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 72
3.7 Analytical Approaches for the Investigation of Patterns:
Speed of Travelling Waves . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 74
3.8 Numerical Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 77
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 78
xi
xii Contents
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 275
Chapter 1
Introduction
Aggregation and traffic-like movement are two of the most common collective
behaviours observed in animal and human communities, as well as in some cell
populations [1–3]. These two behaviours can lead to the formation of a large variety
of complex spatial and spatiotemporal group patterns in various organisms. Milling
schools of fish [4], stationary aggregations formed by resting animals, zigzagging
flocks of birds [5], pedestrian traffic jams [6], bi-directional lanes in ants traffic
[7], or rippling waves in Myxobacteria swarms [8] are only a few of the observed
patterns. The interest in these aggregative and movement behaviours was triggered
by the desire and necessity to understand and control the resulting group patterns.
For example, understanding the formation and movement of schools of fish and
swarms of insects is important to control insect outbreaks [9], or to establish
fishing strategies [10]. Understanding pedestrian movement is useful for improving
the architectural design of buildings, train stations, stadiums, or airport terminal
to increase their efficiency and safety[11–14]. In cell biology, understanding the
collective movement of cells is important in morphogenesis (tissue patterning) and
cancer [15, 16].
There are two types of aggregative and movement behaviours displayed by
organisms: aggregation and movement in response to external factors (e.g., exter-
nal chemicals), and aggregation and movement as a result of interactions with
conspecifics (self-organised behaviours). In this study we focus only on the self-
organised behaviours, and review some classes of mathematical models derived
to investigate them. For more general reviews of mathematical models describing
biological pattern formation (including patterns in response to chemotaxis) we refer
the reader to [17–20].
The mathematical models derived to investigate the self-aggregation patterns
observed in nature aim to propose plausible hypotheses regarding the biological
Mesoscopic level
u(x,t;v)
Fig. 1.1 Caricature description of various scales at which self-organised behaviours can occur
dynamics):
• Individual-based models (or agent-based models): models describing the
behaviour of a number of entities (individuals/cells) with specific characteristics
(e.g., spatial position, velocity) that are tracked through time. These individual
characteristics can change through time in response to some well-defined rules.
• Kinetic models: continuum models that describe the evolution of a probability
distribution of individuals/cells in a population, in a phase space generally
described by the position, time and velocity (orientation) of individuals. In addi-
tion to individuals/cells velocity (or orientation), other specific characteristics
can be incorporated into these models: from individuals’ age and size [23], to the
activation level of cells [24], or the concentration of molecules inside cells or on
cells’ surface [25].
• Macroscopic parabolic and hyperbolic models: continuum models that describe
the spatial distribution of the whole population density (where all individual
characteristics are averaged).
The transitions from microscopic to kinetic and further to macroscopic models
involve different technical approximations, which come with analytical or computa-
tional drawbacks. In particular, the structure of these different types of models leads
to their investigation either from a computational perspective, or from an analytical
perspective (with relatively few models being investigated both analytically and
numerically):
• Computational approaches: the focus is on numerical simulations to repro-
duce and compare the simulated solution patterns with the available data (to
ascertaining whether the assumptions incorporated into the models are enough
to explain the experimentally-observed aggregation patterns). This approach is
commonly found in individual-based models for aggregations (i.e., models that
contain a set of decision rules that govern the movement of individuals, and
in particular their spatial position, speed and turning behaviours) [26–33]. In
spite of the complex group patterns displayed by the individual-based models
(e.g., swarms, tori, polarized groups [26]), the lack of analytical techniques to
investigate them causes difficulties in understanding some of these patterns [34].
In this case, numerical simulations are sometimes the only option to investigate
the behaviour of the models. We note that this approach is also very common
in some nonlinear continuum models for collective behaviours in cells [35, 36],
where the complexity of equations makes it difficult to investigate analytically
the resulting patterns.
• Analytical approaches: the focus is on using existence and uniqueness results to
show that the models can exhibit particular types of solutions, such as stationary
steady-state aggregations [37]. Homogenisation results are also used to try to
connect the dynamics at the micro-scale and macro-scale levels [38]. Such
analytical approaches are very common for continuum models (i.e., models
that describe the evolution of the density of individuals), and in particular for
kinetic models [39–44]. Since the numerical simulations for many of these
4 1 Introduction
kinetic models are too time consuming, the analytical results can offer a quicker
understanding of some dynamics of the models.
Throughout this monograph we will discuss the contribution of both approaches to
our understanding of pattern formation in biological self-organised aggregations. In
particular, we will emphasise the necessity of combining these two approaches to
obtain a complete description of the patterns that can be exhibited by the models.
Since the majority of individual-based models are investigated mainly numerically
(due to a lack of analytical techniques to investigate the existence and structure
of patterns obtained numerically, which might impede the discovery of some
new possible patterns), we decided to focus this monograph only on kinetic and
macroscopic models. However, before we discuss these models, we will briefly
present some individual-based models that are the starting point of kinetic and
hyperbolic models for aggregation and movement.
1. Vicsek Model. Vicsek et al. [28] introduced the following model for the
movement of N self-propelled particles with constant velocity. Each particle
i = 1..N aligns with its neighbours up to a random noise:
Here, xi (t) ∈ R2 denotes the position vector of the i-th particle at time
t ∈ R+ , vi (t) ∈ R2 is the velocity vector of the i-th particle at time t, and
θi (t) ∈ [−π, π] is the angle defining the direction of the velocity. (Note that
throughout this monograph we use the notation R+ = {x ∈ R|x ≥ 0}.)
Moreover, θi (t)r denotes the average direction of the velocities of particles
positioned within a circle of radius r around the reference particle xi (t) (and
θi (t)r = arctan[sin(θi (t))r /cos(θi (t))r ]). Also, vo ∈ R is the magnitude of
particles velocities. Finally, the term Δθ denotes a random noise (i.e., a random
number chosen with a uniform probability from an interval [−η/2, η/2]) [28].
The authors showed that, when varying the noise level η or the number
of particles N, this model exhibits transitions from a disordered behaviour
(characterised by zero averaged velocity) to an ordered behaviour (characterised
by nonzero average velocity), through spontaneous symmetry breaking of the
rotational symmetry [28]. To characterise the overall behaviour of the collective
of particles, the authors introduced an order parameter [28]:
1
N
φ= | vi |. (1.2)
Nvo
i=1
γ
N
vi (t + Δt) = vi (t) + aij vj (t) − vi (t) . (1.3b)
N
j =1
1
aij := β . (1.4)
1 + |xi − xj |2
Parameter N gives the total number of particles in the system, while parameter γ
gives the strength of the interactions. Parameter β describes the rate of decay of
the influence between particles, as they move away from each other. The authors
then found conditions on the parameter β that ensured that all particles in the
aggregation will converge to a common velocity [53]. In particular, for
1 1
Γ (x(t)) = |xi − xj |2 , Λ(v(t)) = |vi − vj |2 , (1.5)
2 2
i=j i=j
the authors show that there exists a constant B0 such that Γ (x(t)) ≤ B0 , for all
t ∈ R, t > 0. Also, Λ(v(t)) → 0 as t → ∞, and xi − xj → x̂ij (where x̂ij
is a limit vector). When β < 1/2 the convergence to a flock with a common
velocity v̂ is always guaranteed. However, for β > 1/2, the authors showed that
convergence happens only for some initial conditions for birds positions (xi (0))
and velocities (vi (0)).
We note that while the Vicsek model (1.1) includes noise, the original Cucker-
Smale model (1.3) does not include any noise. (However, uniform noise could
be easily added to the velocity equation, as shown in [63].) Moreover, while
the particles in Vicsek’s model interact within well-defined spatial ranges (the
authors looked at the velocities of the neighbouring particles within a circle
of radius r around a reference particle i), the dynamics of the Cucker-Smale
model occurs over a very long spatial range (with β controlling the decay of
the interactions over this range). As shown in [53], one could slightly modify the
Cucker-Smale model in the spirit of Vicsek’s model by assuming that the velocity
1.1 Modelling Self-organised Aggregation and Movement 7
γ
N
vi (t + Δt) = vi (t) + aij (|xi − xj |2 ) vj (t) − vi (t) , (1.6)
N
j =1
with
2 1, if |xi − xj |2 ≤ r 2 ,
aij (|xi − xj | ) = (1.7)
0, otherwise,
where r the interaction distance between individuals. However, in this case the
change in the angular velocity is abrupt, which is in contrast with the continuous
change in the velocity for the Cucker-Smale model.
The weights aij are often referred to as “communication rates”. However,
these rates, which are usually distance-dependent, are different from the com-
munication mechanisms we will refer to in Chap. 5. Haskovec [64] introduced
an alternative derivation of these rates in terms of the topological interactions
between individuals:
N
K xi − xk
aij = , with gij = χ , (1.8)
β |xi − xj |
1 + gij2 k=1
with χ the characteristic function of the open unit ball in Rn . Here, the author
assumed that the important quantity is not the metric distance between two
particles i and j , but rather how many intermediate particles k separate them.
The discussion on the appropriateness of using metric-distance models (in
which individuals interact with all neighbours within a certain distance) versus
topological-distance models (in which individuals interact only with a fixed
number of neighbours) has been fuelled in the past years by various empirical
evidence which, in the end, seems to support both types of model assumptions.
Topological interactions can explain the dynamics of starling flocks [33], while
metric-distance interactions (given by the interaction zones) can explain the
dynamics of fish shoals [30] or flocking surf scoters [31]. We will return to this
discussion in Chap. 5, in the context of communication mechanisms.
3. D’Orsogna Model. D’Orsogna et al. [60] and Chuang et al. [51] developed an
individual-based model which assumed that the velocities of particles depend on
an attractive-repulsive potential:
dxi
= vi , (1.9a)
dt
dvi 1
mi = (α − β|vi |2 )vi − ∇U (|xi − xj |), 1 ≤ i ≤ N. (1.9b)
dt N
i=j
8 1 Introduction
with
dvi 1
mi = (α − β|vi |2 )vi − ∇Ur (|xi − xj |)
dt N
i=j
1 xi − xj
− aij , 1 ≤ i ≤ N. (1.11)
N |xi − xj |
i=j
where Ur is the repulsive component of the potential U , and the attractive force
depends on the relative separation aij of the i-th and j -th individuals/particles.
The introduction of this particular attractive interaction has been shown to lead
to chaotic movement [64].
To conclude the discussion of these models with attractive-repulsive veloci-
ties, we note that phase transitions between different types of patterns have been
observed also in model (1.9), as the ratios lr / la and Cr /Ca are varied. Moreover,
in this class of models the transitions between certain patterns (e.g., from single
to double mills, or from rigid-body rotations to single mills) can be either gradual
or abrupt processes (see Fig. 8(b) in [51]), which is in contrast with the phase
transitions observed in Vicsek’s model [51].
4. Generalised Langevin Model. Chavanis and Sire [65] started with a generalised
Langevin (individual-based) model, and further used it to derive a class of general
kinetic and hydrodynamic/hyperbolic models for chemotactic aggregations.
1.1 Modelling Self-organised Aggregation and Movement 9
Since we will discuss these kinetic and hydrodynamic models in Chap. 6 in the
context of stochastic models for collective movement, here we briefly introduce
the generalised Langevin model for the collective movement of N bacteria in
response to a chemical c(x, t):
dxi
= vi , i = 1, . . . N, (1.12a)
dt
dvi √
= −ξ vi + ∇c + 2DRi (t), (1.12b)
dt
∂c N
= −kc + Dc Δc + h δ(x − xi (t)). (1.12c)
∂t
i=1
10 1 Introduction
dxi
= vi , i = 1, . . . N, (1.13a)
dt
dvi 1 √
= −ξ vi − ∇U (x1 , . . . xN ) + 2DRi (t), (1.13b)
dt m
with
U (x1 , . . . , xN ) = m2 ULR (|xi − xj |) + m2 USR (|xi − xj |)
i<j i<j
+m Uext (xi ), (1.14)
i
λ+i λ+i
unidirectional
reception M3
(e.g., visual xi −si xi xi +s i xi −si xi x i+s i
signals)
ZLa ZRa
Interaction Ranges: xi −si xi −si ... xi −si ... x i ... xi +si ... xi +s i xi +s i
(1D domain)
for individual/cell "i" ZLa ZLal ZLr ZR
r
R
Zal ZaR
particle/cell/individual is
±
ya;i = ±qa Ka (|xi − xj |) ∓ qa Ka (|xi − xj |), (1.16)
j ∈Za;i
R j ∈Za;i
L
with Ka (·) an interaction kernel describing the width of the attractive interaction
zone (e.g., a Gaussian kernel in [67], or any of the kernels summarised in
Fig. 5.2). Similar interaction forces are derived for repulsion and alignment.
Following the approach in [68, 69] (for hyperbolic/kinetic models), the authors
in [67] have defined different types of nonlocal communication mechanisms,
which influence the way social interactions terms yi± affect the turning rates
λ±i ; see Fig. 1.3. A summary of the different yi terms, corresponding to the
communication mechanisms in Fig. 1.3 is shown in Table 1.1. (We will discuss
the importance of inter-individual communication on collective behaviours in
Sect. 1.2.)
In terms of pattern formation, Zmurchok and de Vries [67] have shown that
these nonlocal individual-based models can reproduce the majority of the patterns
Table 1.1 The nonlocal terms used to describe the social interactions in [67]
± ± ±
Model Social interactions: repulsion (yr,i ), attraction (ya,i ), alignment (yal,i )
±
M1 yr,a;i = ±qr,a j ∈Zr,a;i
R Kr,a (dij ) ∓ qr,a j ∈Zr,a;i
L Kr,a (dij )
±
yal;i = ±qal j ∈Zal;i
R , v <0
j
Kal (dij ) ∓ qal j ∈Zal;i
L ,v
j >0
Kal (dij )
±
M2 yr,a;i = ±qr,a j ∈Zr,a;i
R Kr,a (dij ) ∓ qr,a j ∈Zr,a;i
L Kr,a (dij )
±
yal;i = ±qal j ∈Zal;i
R , v <0
j
Kal (dij ) ± qal j ∈Zal;i
L , v <0
j
Kal (dij )
∓qal j ∈Zal;i
R , v >0
j
Kal (dij ) ∓ qal j ∈Zal;i
L Kal (dij )
+ −
M3 yr,a;i = qr,a j ∈Zr,a;i
R Kr,a (dij ), yr,a;i = qr,a
j ∈Zr,a;i
L Kr,a (dij )
+
yal;i = qal j ∈ZR , vj <0 Kal (dij ) − qal j ∈ZR , vj >0 Kal (dij ),
al;i al;i
−
yal;i = qal j ∈ZL , vj >0 Kal (dij ) − qal j ∈ZL , vj <0 Kal (dij ),
al;i al;i
±
M4 yr,a;i = ±qr,a j ∈ZR , vj <0 Kr,a (dij ) ∓ qr,a j ∈ZL , vj >0 Kr,a (dij )
r,a;i r,a;i
±
yal;i = ±qal j ∈ZR , vj <0 Kal (dij ) ∓ qal j ∈ZL , vj >0 Kal (dij )
al;i al;i
+ −
M5 yr,a;i = qr,a j ∈ZR , vj <0 Kr,a (dij ), yr,a;i = qr,a;i j ∈ZL , vj >0 Kr,a (dij )
r,a;i al;i
+ −
yal;i = qal j ∈ZR , vj <0 Kal (dij ), yal;i = qal j ∈ZL , vj >0 Kal (dij )
al;i al;i
These terms are the translations of the diagrams from Fig. 1.3 into mathematical equations, after
summing up the information received from all neighbours (si ∈ (0, ∞)). For each of the five
± ±
models, the equations for yr;i and ya;i are almost identical (since the effect of the repulsive and
attractive interactions is similar, but leads to opposite outcomes). Here qa , qr , and qal describe the
strength of the attraction, repulsion, and alignment interactions between the individual at xi at its
neighbours at xj , where dij = |xi − xj |. Kernels Ka (dij ), Kr (dij ) and Kal (dij ) give the width of
the attractive (ZaR,L ), repulsive (ZrR,L ) and alignment (Zal
R,L
) interaction zones. Since the direction
of the neighbours j is important for social interactions via communication, the sums distinguish
between individuals with vj > 0 and vj < 0
1.1 Modelling Self-organised Aggregation and Movement 13
t
2400 2400 2400
2200 2200 2200
2000 2000 2000
0 5 10 0 5 10 0 5 10
x x x
(d) (e) (f)
3000 3000 3000
2800 2800 2800
2600 2600 2600
t
t
2400 2400 2400
2200 2200 2200
2000 2000 2000
0 5 10 0 5 10 0 5 10
x x x
(g) (h) (i)
3000 3000 3000
2800 2800 2800
(e) (f)
space (y)
Fig. 1.5 Caricature description of the patterns that could be generated by two-population
individual-based models. (a) stationary aggregations with complete engulfment; (b) moving
aggregations (as given by the direction of the arrows) with complete engulfment; (c) moving aggre-
gations with partial engulfment; (d) stationary segregated aggregations; (e) moving segregating
aggregations; (f) segregated line formation
1.1 Modelling Self-organised Aggregation and Movement 15
(as displayed by both discrete and continuum models; see also the discussion in
Sect. 6.10).
• Pineda & Eftimie Model. Pineda and Eftimie [72] introduced an individual-
based model to describe the self-organised dynamics of two cell populations
which interact not only through spatially-dependent repulsive social forces, but
also through a chemical that can be secreted by the cells themselves or can
be secreted externally (which leads to an externally imposed stable chemical
gradient). Thus, assuming that a population of N cells is formed of two
subpopulations NA and NB of cells with different speeds (vi ) and different
chemotactic sensitivities (χi ), then the equations for the evolution of the two
populations (characterised by position xi = (xi , yi ) ∈ R2 and velocity vi ∈ R2 )
are as follows:
with
cos θi (t)
vi (t) = vi ei (t) = vi , θi ∈ [−π, π], (1.18)
sin θi (t)
and
⎧ ⎫
⎨ ∇cT (t)
N ⎬
θi (t + Δt) = arg χi + mij Fij (t) + αi σ i (t) (1.19)
⎩ ||∇cT (t)|| ⎭
j =1,i=j
N
∇cI (xi ) = −p KI (||xi − xj ||/ l)xij ,
j =1,j =i
√ √
μE e− μE DE y + e μE DE y
∇cE (y) = −cm √ √ y. (1.20)
DE e− μE DE y + e μE DE y
diffusion rate of the chemical cE , μE is the degradation rate of this chemical, and
y is a unitary vector pointing in the y-direction.
The second term in Eq. (1.19) describes the repulsive interactions between
cells i and j : Fij = fij xij , with
0, if xij (t) > 2r0 ,
fij = xij (1.21)
1 − 2r0 , if 0 ≤ xij (t) ≤ 2r0 ,
where xij (t) = ||xi (t) − xj (t)|| is the distance between two cells, and 2r0 is
the interaction distance between two cells. The parameter mij (from Eq. (1.19))
controls the magnitude of the repulsive cell-cell interactions.
Finally, the last term in Eq. (1.19) describes the noise affecting cells orien-
tation: αi is the noise intensity, while σ i = (cos ξi (t), sin ξi (t)) is a random
uniformly oriented unitary vector.
Numerical simulations with this individual-based model (where individual
cell dynamics was coupled with continuous chemical gradients) have shown: (1)
moving aggregations with complete engulfment of one population (with stronger
chemotactic sensitivity) by the other population (as in Fig. 1.5b); (2) moving
aggregations with partial engulfment of one population by the other population
(as in Fig. 1.5c); (3) segregated moving cell aggregations (as in Fig. 1.5e); (4) lane
formation, with the two lanes—corresponding to the two populations—moving
towards the external chemical gradient (as in Fig. 1.5f). We emphasise that while
all aggregation patterns in [72] are moving because cells are attracted to the
external chemical gradient cE , it is possible to obtain also stationary aggregations
(as those summarised in Fig. 1.5) for different parameter values. We will return to
these patterns in Chap. 6, in the context of patterns generated by two-population
kinetic models.
The models in [65] and [72] could be seen as connecting micro-scale dynamics
(as represented by the movement of individual cells/bacteria) to macro-scale
dynamics (as represented by the time-evolution of the diffusive chemicals). Another
approach that connects the micro- and macro-scale models is based on the derivation
of microscopic models on a lattice with cells of length Δx. The models (sometimes
called “master equations”) describe the probability p(x) of a particle/pedestrian,
initially positioned in cell x ∈ R, to move right (to x + Δx) or left (to x − Δx). (In
2D the particles can move also up and down.) In the limit Δx → 0, this microscopic
model becomes a macroscopic model for the density u(x) of particles/pedestrians at
x (where x is now a continuous space variable). This approach has been considered
for the derivation of some local and nonlocal hyperbolic systems with constant speed
[73–75], or for the derivation of local hyperbolic models with density-dependent
speed [76]. We will discuss these models in more detail in Chaps. 4 and 5.
Remark 1.1 While many individual-based models report results that are visually
similar to experimentally-observed collective migration (e.g., see [59, 77]), a
quantitative comparison between these numerical results and experimental data
1.1 Modelling Self-organised Aggregation and Movement 17
is a very tedious process. In this case, one needs to perform a large number
of simulations to obtain appropriate statistics (suitable for comparison with data
statistics). For this reason, the majority of individual-based models in the literature
ignore the quantitative comparison step.
Editor: S. F. Harmer
Sir A. E. Shipley
Language: English
Credits: Keith Edkins, Peter Becker and the Online Distributed Proofreading
Team at https://www.pgdp.net (This file was produced from images
generously made available by The Internet Archive)
EDITED BY
AND
VOLUME II
NEMERTINES
By Miss L. Sheldon, Newnham College, Cambridge
ROTIFERS
By Marcus Hartog, M.A., Trinity College, Cambridge (D.Sc. Lond.), Professor of Natural
History in the Queen's College, Cork
POLYCHAET WORMS
By W. Blaxland Benham, D.Sc. (Lond.), Hon. M.A. (Oxon.), Aldrichian Demonstrator of
Comparative Anatomy in the University of Oxford
POLYZOA
By S. F. Harmer, M.A., Fellow of King's College, Cambridge
London
MACMILLAN AND CO., Limited
NEW YORK: THE MACMILLAN COMPANY
1901
PAGE
Scheme of the Classification adopted in this Book ix
NEMERTINEA
CHAPTER V
NEMERTINEA
Introductory—External Characters—Anatomy—Classification—
Development—Habits—Regeneration—Breeding—
Geographical Distribution—Land, Fresh-Water, and Parasitic
Forms—Affinities 99
POLYZOA
CHAPTER XVII
POLYZOA
Introduction—General Characters and Terminology—Brown
Bodies—History—Outlines of Classification—Marine Polyzoa
—Occurrence—Forms of Colony and of Zooecia—Ovicells—
Avicularia—Vibracula—Entoprocta 465
CHAPTER XVIII
POLYZOA—continued
Fresh-water Polyzoa—Phylactolaemata—Occurrence—Structure
of Cristatella—Division of Colony—Movements of Colony—
Retraction And Protrusion of Polypides in Polyzoa—
Statoblasts—Table for Determination of Genera of Fresh-
water Polyzoa—Reproductive Processes of Polyzoa—
Development—Affinities—Metamorphosis—Budding 492
CHAPTER XIX
POLYZOA—continued
Classification—Geographical Distribution—Palaeontology—
Methods for the Examination of Specific Characters—
Terminology—Key for the Determination of the Genera of
British Marine Polyzoa 515
Addendum to Chaetognatha 534
Index 535
SCHEME OF THE CLASSIFICATION ADOPTED IN THIS BOOK
PLATYHELMINTHES (p. 3)
Family.
TURBELLARIA Planoceridae
(p. 3) (p. 19).
Leptoplanidae
Acotylea (p. 16) (p. 19).
Cestoplanidae
(p. 19).
Polycladida
Enantiidae (p. 19).
(p. 7)
Anonymidae (p. 19)
Pseudoceridae
(p. 19).
Cotylea
Euryleptidae (p. 19).
Prosthiostomatidae
(p. 19).
Paludicola
Planariidae (p. 42).
(p. 30)
Procerodidae
Maricola (p. 42).
Tricladida (pp. 30, 32) = Gundidae.
(p. 30) Bdellouridae (p. 42).
Bipaliidae (p. 42).
Terricola Geoplanidae (p. 42).
(pp. 30, 33) Rhynchodemidae
(p. 42).
Rhabdocoelida Proporidae (p. 49).
(p. 42) Acoela (p. 42) Aphanostomatidae
(p. 49).
Macrostomatidae
(p. 49).
Microstomatidae
(p. 49).
Prorhynchidae
Rhabdocoela (p. 49).
(p. 43) Mesostomatidae
(p. 49).
Proboscidae (p. 49).
Vorticidae (p. 50).
Solenopharyngidae
(p. 50).
Alloeocoela Plagiostomatidae
(p. 43) (p. 50).
Bothrioplanidae
(p. 50).
Monotidae (p. 50).
Temnocephalidae
(pp. 53, 73).
Tristomatidae
(pp. 53, 73).
Monogenea (pp. 5, 52)
Polystomatidae
= Heterocotylea + Aspidocotylea
(pp. 53, 73).
(p. 73)
Gyrodactylidae
(pp. 53, 61).
Aspidobothridae
(p. 73).
TREMATODA Holostomatidae
(pp. 3, 51) (p. 73).
Amphistomatidae
(p. 73).
Distomatidae
Digenea (pp. 5, 52) = Malacocotylea (p. 73).
(p. 73) Gasterostomatidae
(p. 73).
Didymozoontidae
(p. 73).
Monostomatidae
(p. 73).
Cestodariidae
= Monozoa (p. 91).
Bothriocephalidae
(p. 91).
CESTODA (pp. 3, 74) Tetrarhynchidae
(p. 91).
Tetraphyllidae
(p. 91).
Taeniidae (p. 91).
MESOZOA
Dicyemidae (p. 93).
MESOZOA (pp. 3, 92) Orthonectida
(p. 94).
NEMERTINEA (p. 99)
HOPLONEMERTEA (p. 110) = Metanemertini (p. 112).
SCHIZONEMERTEA (p. 111) = Heteronemertini (ex parte) (p. 113).
PALAEONEMERTEA (p. 111) = Protonemertini (p. 112). + Mesonemertini
(p. 112). + Heteronemertini (ex parte) (p. 113).
Echinorhynchidae
(p. 182)
Gigantorhynchidae
(p. 183).
ACANTHOCEPHALA (pp. 123, 174)
Neorhynchidae
(p. 184).
Arhynchidae
(p. 185).
Melicertidae
(p. 221).
MELICERTACEAE (p. 221)
Trochosphaeridae
(p. 221).
Philodinidae
BDELLOIDA (p. 222)
(p. 222).
Asplanchnidae
ASPLANCHNACEAE (p. 222)
(p. 223).
Pedalionidae
SCIRTOPODA (p. 223)
(p. 223).
Microcodonidae
(p. 224).
Rhinopidae (p. 224).
Hydatinidae
(p. 224).
Synchaetidae
Illoricata (p. 223)
(p. 224).
Notommatidae
(p. 224).
Drilophagidae
(p. 224).
Triarthridae (p. 224).
Rattulidae (p. 225).
PLOIMA (p. 223)
Dinocharididae
(p. 225).
Salpinidae (p. 225).
Euchlanididae
(p. 225).
Cathypnidae
Loricata (p. 224)
(p. 225).
Coluridae (p. 225).
Pterodinidae
(p. 225).
Brachionidae
(p. 225).
Anuraeidae (p. 225).
SEISONACEAE (p. 225) Seisonidae (p. 226).
GASTROTRICHA
GASTROTRICHA Euichthydina (p. 235)
(p. 231). Apodina (p. 235)
Gnathobdellidae
(p. 407).
GNATHOBDELLAE (p. 407)
Herpobdellidae
(p. 407).
BY
CHAPTER I
TURBELLARIA
The Trematodes[2] may be divided into those living on the outer surface of various
aquatic animals, usually fish (Ectoparasites); and those which penetrate more or
less deeply into the alimentary canal or the associated organs of the host
(Endoparasites). They are oval, flattened Platyhelminthes ranging from a
microscopic size to a length of three feet (Nematobothrium, Fig. 22), and are
provided with organs of adhesion by which they cling to the outer surface, or to the
interior, of the animals they inhabit. Trematodes occur parasitically in all groups of
Vertebrates, but, with the exception of the liver-flukes of the sheep (Distomum
hepaticum and D. magnum), and of Bilharzia haematobia found in man (in the
blood-vessels of the urinary bladder) over the greater part of Africa, their attacks
are not usually of a serious nature. Ectoparasitic Trematodes are Monogenetic;
that is, their larvae grow up directly into mature forms. The Endoparasitic species,
however, are usually Digenetic. Their larvae enter an Invertebrate and produce a
new generation of different larvae, and these another. The last are immature
flukes. They enter a second host, which is swallowed by the final Vertebrate host in
which they become mature.