Hyperbolic and Kinetic Models for Self

organised Biological Aggregations A

Modelling and Pattern Formation
Approach Raluca Eftimie
Visit to download the full and correct content document:
https://textbookfull.com/product/hyperbolic-and-kinetic-models-for-self-organised-biolo
gical-aggregations-a-modelling-and-pattern-formation-approach-raluca-eftimie/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Python Debugging for AI, Machine Learning, and Cloud

Computing: A Pattern-Oriented Approach 1st Edition
Vostokov

https://textbookfull.com/product/python-debugging-for-ai-machine-
learning-and-cloud-computing-a-pattern-oriented-approach-1st-
edition-vostokov/

From Mafia to Organised Crime: A Comparative Analysis

of Policing Models 1st Edition Dr. Anna Sergi (Auth.)

https://textbookfull.com/product/from-mafia-to-organised-crime-a-
comparative-analysis-of-policing-models-1st-edition-dr-anna-
sergi-auth/

Formation of KNbO3 Thin Films for Self Powered ReRAM

Devices and Artificial Synapses Tae-Ho Lee

https://textbookfull.com/product/formation-of-knbo3-thin-films-
for-self-powered-reram-devices-and-artificial-synapses-tae-ho-
lee/

Feature Engineering and Selection A Practical Approach

for Predictive Models 1st Edition Max Kuhn

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/

Computed Tomography of the Lung A Pattern Approach 2nd

Edition Johny A. Verschakelen

https://textbookfull.com/product/computed-tomography-of-the-lung-
a-pattern-approach-2nd-edition-johny-a-verschakelen/

Radiology Illustrated: Chest Radiology: Pattern

Approach for Lung Imaging 2nd Edition Lee

https://textbookfull.com/product/radiology-illustrated-chest-
radiology-pattern-approach-for-lung-imaging-2nd-edition-lee/

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/

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).

More information about this series at http://www.springer.com/series/304

Raluca Eftimie

Hyperbolic and Kinetic

Models for Self-organised
Biological Aggregations
A Modelling and Pattern Formation Approach

123
Raluca Eftimie
Division of Mathematics
University of Dundee
Dundee, UK

ISSN 0075-8434 ISSN 1617-9692 (electronic)

Lecture Notes in Mathematics
ISSN 2524-6771 ISSN 2524-678X (electronic)
Mathematical Biosciences Subseries
ISBN 978-3-030-02585-4 ISBN 978-3-030-02586-1 (eBook)
https://doi.org/10.1007/978-3-030-02586-1

Library of Congress Control Number: 2018963056

Mathematics Subject Classification (2010): 92-01, 92-02, 92C15, 92D50, 35C07, 35Lxx, 35Q20, 35Q92,
35R09, 35R60, 37G40, 58J55, 65Nxx

© Springer Nature Switzerland AG 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

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

Self-organised biological aggregations (i.e., aggregations that form in the absence of

a leader or external factors) have caught the interest and imagination of scientists for
thousands of years. For example, Pliny the Elder, in his book The Natural History
(published around 77–79 AD) [1], discussed the migration and movement of various
flocks of birds, from thrushes and blackbirds to starlings and swallows. About
starlings, he noted that “it is a peculiarity of the starling to fly in troops, as it were,
and then to wheel round in a globular mass like a ball, the central troop acting as a
pivot for the rest”. In regard to swallows, Pliny the Elder remarked that they are “the
only birds that have a sinuous flight of remarkable velocity” [1]. These comments
emphasise humanity’s long-term fascination with the spatial and spatio-temporal
patterns displayed by various animal aggregations (including flocks of birds).
Over the last 60 years or so, researchers have used mathematical models to
identify the biological mechanisms that could explain the formation and structure
of these animal group patterns. One of the first studies in this area was published
by Breder [2], who used algebraic equations for the repulsive-attractive forces
among individuals, to understand “the basic nature of influences at work in a school
of fishes as well as in other less compact aggregations”. In his pioneering work,
Breder proposed the idea that there are attraction and repulsion forces between fish,
which vary with the distance between them and those forces are likely mediated
by different sensory mechanisms. Breder [2] identified vision to be the main factor
involved in the attraction of fish towards each other and suggested that repulsion
could be caused not only by vision but also by water movement or sound.
Currently, most of mathematical models for self-organised aggregations assume,
one way or another, that the attractive-repulsive interactions are the basic mech-
anism behind the formation and persistence of biological aggregations. (Here,
I define self-organised aggregations as being those biological aggregations that
form in the absence of a leader or some external stimulus.) In general, these
social interactions are nonlocal, with repulsion acting on short distances and
attraction acting on large spatial distances. Thus, many mathematical models for
self-organised biological aggregations are nonlocal. Nevertheless, the incorporation
into the mathematical models of these social interactions alone cannot explain all

vii
viii Preface

complex patterns observed empirically in cell, bacterial and animal aggregations.

This led researchers to consider another social interaction: alignment/polarisation.
However, alignment behaviours cannot be properly described with the help of
parabolic-type partial differential equations (which focus on random movements).
A more natural approach for modelling alignment behaviours sees the use of
hyperbolic and kinetic transport models.
The purpose of this monograph is to introduce this research area, of mathematical
approaches for the investigation of spatial and spatio-temporal patterns displayed by
self-organised biological aggregations, to students and researchers not familiar with
the topic. To this end, I consider a step-by-step approach to describe various 1D
and 2D local hyperbolic and kinetic models (where interactions depend only on
the local density of neighbours), as well as nonlocal models (where individuals can
perceive, via different sensory mechanisms, their conspecifics positioned further
away). I discuss the patterns obtained numerically with these models, as well as
other patterns that have been shown to exist or not with the help of analytical
methods. For completeness, I also give a brief overview of the most common
analytical approaches used to investigate the dynamics of hyperbolic and kinetic
models. In addition, I discuss briefly a variety of numerical schemes developed to
approximate the solutions of different hyperbolic and kinetic models (mainly related
to problems in physics, but which can be considered also in biological contexts).
The complexity and variety of these hyperbolic and kinetic models makes
it difficult to include here all types of models existent in the literature (and
all analytical and numerical approaches developed to investigate these models).
Moreover, because the investigation of collective aggregations and movement is
currently one of the most active research areas in mathematical biology, more and
more models are developed every month. Since one needs to stop somewhere, I
tried to focus on models that either introduced a new idea in terms of modelling
self-organised aggregations or used particular analytical and numerical techniques
to investigate the formation of patterns. However, there is a feeling that many
modelling/analytical/numerical aspects should have been presented in more detail.
Ultimately, I hope that this monograph will offer a first overview into the
use of kinetic and hyperbolic models to reproduce and investigate stationary and
moving biological aggregations. Moreover, I hope that researchers interested in
analytical and numerical approaches for hyperbolic and kinetic models (that have
been applied so far mainly to problems in physics and engineering) will become
aware of the complexity of phenomena in biology and the numerous open analytical
and numerical problems associated with the models for self-organised biological
aggregations.
This monograph is the result of multiple research collaborations (over the past
15 years) on various topics related to pattern formation in ecological and biological
systems and discussions with colleagues and mentors. I am particularly grateful to
my PhD supervisors, Prof. Mark A. Lewis and Prof. Gerda de Vries, as well as to
Prof. Frithjof Lutscher (with whom I collaborated at the beginning of my PhD),
who introduced me to the use of hyperbolic systems to describe 1D movement in
biological/ecological aggregations and guided my first steps in the analytical and
Preface ix

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.

Dundee, UK Raluca Eftimie

October 2018

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

4 Local Hyperbolic/Kinetic Systems in 1D . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 81

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 81
4.2 Derivation of Local Models in 1D . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 83
4.3 Density-Dependent Turning Rates . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 87
4.4 Analytical Approaches for the Investigation of Patterns:
Stability of Homogeneous States and Travelling Waves for
Models with Density-Dependent Turning Rates .. . . . . . . . . . . . . . . . . . . . 89
4.5 Density-Dependent Speeds .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 92
4.6 Models that Include Reaction Terms . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 96
4.7 Traffic Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 99
4.7.1 First-Order Models . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 99
4.7.2 Second-Order Models .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 100
4.8 Multiple Population Models . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 102
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 104
5 Nonlocal Hyperbolic Models in 1D. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 107
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 107
5.2 One-Equation Nonlocal Models with Density-Dependent
Speeds .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 109
5.3 Derivation of Nonlocal Hyperbolic Systems . . . . .. . . . . . . . . . . . . . . . . . . . 113
5.4 Alignment Models .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 116
5.5 Repulsive, Attractive, and Alignment Models . . .. . . . . . . . . . . . . . . . . . . . 117
5.6 Multiple Populations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 127
5.7 Analytical Approaches for the Investigation of Patterns: Loss
of Patterns in the Parabolic Limit . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 137
5.8 Systems with Density-Dependent Speeds . . . . . . . .. . . . . . . . . . . . . . . . . . . . 140
5.9 Systems Incorporating Environmental Stochasticity .. . . . . . . . . . . . . . . . 145
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 149
6 Multi-Dimensional Transport Equations . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 153
6.1 General Description of 2D and 3D Models. . . . . . .. . . . . . . . . . . . . . . . . . . . 153
6.2 Local Advective Models in 2D. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 156
6.3 Nonlocal Advective Models in 2D . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 158
6.4 Velocity-Jump Processes . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 160
6.5 Kinetic Models for Active Particles. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 168
6.6 Multiscale Models with Explicit Dynamics for the Microscale
Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 171
6.7 Derivation of Mean-Field Models . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 175
6.8 Analytical Approaches for the Investigation of Patterns:
Hyperbolic (and Parabolic) Scaling .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 180
6.9 Analytical Approaches for the Investigation of Patterns:
Grazing Collision Limit . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 182
6.10 Multiple Population Models with Nonlocal Interactions . . . . . . . . . . . . 186
6.11 Explicit Stochastic Models .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 189
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 191
Contents xiii

7 Numerical Approaches for Kinetic and Hyperbolic Models . . . . . . . . . . . . 195

7.1 Numerical Schemes for Hyperbolic Models . . . . .. . . . . . . . . . . . . . . . . . . . 196
7.1.1 1D Finite Difference and Finite Volume Methods .. . . . . . . . . . 197
7.1.2 Higher Resolution Schemes for Nonlinear Equations . . . . . . . 202
7.1.3 Multi-Dimensional Finite Difference and Finite
Volume Schemes . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 205
7.2 Numerical Schemes for Higher Dimension Kinetic Models.. . . . . . . . 207
7.2.1 Spectral Methods . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 208
7.2.2 Monte Carlo Methods . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 211
7.2.3 Multi-Scale Stiff Kinetic Problems: Asymptotic
Preserving Methods and Domain
Decomposition Methods . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 213
7.3 Boundary Conditions .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 221
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 224
8 A Few Notions of Stability and Bifurcation Theory . .. . . . . . . . . . . . . . . . . . . . 227
8.1 Basic Notions of Linear Stability Analysis . . . . . . .. . . . . . . . . . . . . . . . . . . . 228
8.1.1 Linear Stability Analysis for ODE Models . . . . . . . . . . . . . . . . . . 228
8.1.2 Linear Stability Analysis for PDE Models .. . . . . . . . . . . . . . . . . . 229
8.2 Basic Notions of Bifurcation Theory . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 233
8.3 Symmetry of Hyperbolic and Kinetic Equations .. . . . . . . . . . . . . . . . . . . . 241
8.4 Compact Operators and the Fredholm Alternative .. . . . . . . . . . . . . . . . . . 245
8.5 Analytical Approaches for the Investigation of Patterns:
Weakly Nonlinear Analysis . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 248
8.6 Centre Manifold Theory .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 255
8.7 Stochastic Bifurcations . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 261
8.8 Bifurcation and Symmetry Theory in the Context
of Hyperbolic/Kinetic Models . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 262
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 263
9 Discussion and Further Open Problems .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 265
9.1 Summary .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 265
9.2 Biological Relevance of Models’ Assumptions and
Generated Numerical Patterns . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 266
9.3 Directions for Future Research. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 268
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 272

Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 275
Chapter 1
Introduction

1.1 Modelling Self-organised Aggregation and Movement

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

© Springer Nature Switzerland AG 2018 1

R. Eftimie, Hyperbolic and Kinetic Models for Self-organised Biological
Aggregations, Lecture Notes in Mathematics 2232,
https://doi.org/10.1007/978-3-030-02586-1_1
2 1 Introduction

Microscopic (individual) level Macroscopic (population) level

vk
vi
xk vm
xi xj u(x,t)
vj xm

xi position of individual "i" density distribution of

vi velocity of individual "i" population "u"

Mesoscopic level

u(x,t;v)

probability density distribution of

population "u", with individual
characteristics "v"
(v velocity, age, activity level, ...)

Fig. 1.1 Caricature description of various scales at which self-organised behaviours can occur

mechanisms at the microscale-level (e.g., speed of animals/cells, turning rate of

animals/cells, presence/absence of attractive-repulsive social interactions, molecular
pathways inside cells that control movement and cell-cell interactions) that can
explain the observed group-level behaviours (e.g., the shape, size, and movement of
animal or cellular aggregation). In Fig. 1.1 we exemplify the relations between the
different scales over which dynamics can occur during self-organised behaviours in
animal and cell communities. Note that for collective cell dynamics, various studies
interpret differently the micro-scale level. For example, in [21] the authors see the
micro-scale level as the level of individual cells, which move as discrete entities
with specific positions and velocities (as in Fig. 1.1), while in [22] the authors see
the microscale level as the level of molecular processes that take place either inside
the cell or on the cell surface (e.g., receptor binding, activation of specific signalling
pathways inside cells). To clarify this issue, throughout this study we will refer to
the cell-level versus molecular-level dynamics (and note that in some studies the
cell-level can be connected to the molecular-level, thus giving rise to mesoscale
models).
Based on these different scales of cell/animal dynamics, the mathematical models
for self-organised biological aggregations can be classified as follows: individual-
based models (for microscale-level dynamics), kinetic models (for mesoscale-level
dynamics) and macroscopic parabolic and hyperbolic models (for macroscale-level
1.1 Modelling Self-organised Aggregation and Movement 3

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.

1.1.1 Individual-Based Models

There is a vast literature of individual-based models that describe self-organised

biological aggregations; see for example [21, 26, 29, 45–59], and the references
therein. Generally, these models are suitable to investigate the dynamics of small
and medium populations. Very large populations, such as tumour aggregations
formed of 109 − 1012 cells, are still difficult to be simulated numerically. The
behaviour of these large populations is usually modelled with the help of continuum
models. To connect the micro- and macro-scales, many studies start from individual-
based models and, in the limit of large numbers, derive equivalent kinetic and
then macroscopic models. However, the original individual-based models and the
limiting continuum models might not always agree in terms of the patterns they
exhibit [51] (since the macroscopic variables are obtained by averaging microscopic
variables that might play important roles in specific dynamics). One such example
is the existence of double milling patterns (see Fig. 1.2b) displayed by individual-
based models [51, 60], and the lack of these patterns in the corresponding continuum
models [44]. The validity of continuum models will be discussed in more detail in
the following sub-section.
Next, we present some individual-based models that are the starting point of
some of the continuum kinetic models that we will investigate in the following chap-
ters. We start with a few individual-based models for homogeneous populations, and
conclude with a model for a heterogeneous population.
1.1 Modelling Self-organised Aggregation and Movement 5

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:

xi (t + Δt) = xi (t) + vi (t)Δt, (1.1a)

 
cos(θi (t))
vi (t + Δt) = vo , (1.1b)
sin(θi (t))
θi (t + Δt) = θi (t)r + Δθ. (1.1c)

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

Here φ is called an order parameter, since it characterises the level of order

(or disorder) in a multi-particles system. Changes in this order parameter, as a
result of changes in various external model parameters (e.g., magnitude of noise),
can give rise to phase transitions, where the particles undergo a change in their
collective behaviour [61]. As discussed in [61], phase transitions can be:
1. discontinuous (or of the first order), when the order parameter that measures
the group structure undergoes a jump between two different values;
2. continuous (or of the second order), when the order parameter changes
quickly its value, but does not undergo a discontinuous jump.
Model (1.1) was shown to exhibit second order phase transitions [28, 61].
We conclude this discussion on phase transitions, by noting that in [62]
the authors mentioned the analogy between phase transitions and bifurcations,
emphasizing that “a phase transition is nothing other than a bifurcation in the
underlying microscopic dynamics”. We will return to this aspect in Chap. 6, in
6 1 Introduction

the context of subcritical and supercritical bifurcations for models described by

partial differential equations.
2. Cucker-Smale Model. Cucker and Smale [53] focused on flocks of birds and
considered Vicsek’s model (1.1) as a starting point for modelling changes in
birds’ positions and velocities. However, these authors assumed that each particle
(i.e., bird in a flock) adjusts its velocity by adding to it a weighted average of the
differences between its velocity and the velocities of its neighbours:

xi (t + Δt) = xi (t) + vi (t)Δt, (1.3a)

γ  
N

vi (t + Δt) = vi (t) + aij vj (t) − vi (t) . (1.3b)
N
j =1

As before, xi , vi ∈ Rn (with n = 3 in [53]) for each particle i = 1, . . . , N.

The weights aij are assumed to depend on the distance between particles:

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

is a function of aij (|xi − xj |2 ) which acts only on finite distances:

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

U (|xi − xj | ) = Cr e−|xi −xj |/ lr − Ca e−|xi −xj |/ la . (1.10)

As before, xi ∈ Rn and vi ∈ Rn describe the position and velocity of particle i

at time t ∈ R+ , mi ∈ R+ is the mass of this particle, α > 0 describes the self-
propulsion strength, while β > 0 is the friction. In the expression of the potential
function U , Cr and Ca represent the magnitudes of the repulsive and attractive
interactions, while lr and la represent the potential ranges for the repulsive and
attractive social interactions. In [60], the authors focus on the H-stability property
of the system, which ensures that the system preserves inter-particle space and
does not collapse as N → ∞. (Note that a system of N particles interacting via
a potential U is called H-stable if there is a constant B > 0 such that U ≥ −NB.
If this property does not hold, then the system is called catastrophic, since the
aggregations collapse to their core.) The authors then go on and investigate the
formation of H-stable and catastrophic patterns, such as clumps (i.e., stationary
pulses), rings (i.e., vortices or mills; single mills and double mills), coherent
flocks (i.e., travelling pulses) or rigid-body rotation states (particles with fixed
positions that rotate around the swarm centre), in the parameter space determined
by the ratios lr / la and Cr /Ca . For a visual description of these patterns see
Fig. 1.2, while for their definitions see Table 1.2.
Haskovec [64] adapted the model (1.9) to incorporate topological-metric
interactions in the equation for the velocity:

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

(a) (b) (c)

space (y)

(d) (e) (f)

space (y)

(g) (h) (i)

space (y)

space (x) space (x) space (x)

Fig. 1.2 Caricature description of various 2D patterns exhibited by one-population individual-

based models: (a) Single rings (mills); (b) Double rings (mills) formed of clockwise and counter-
clockwise rotating rings; (c) Travelling groups (or pulses); (d) Lane patterns; (e) Swarm patterns;
(f) Collapsed swarm patterns; (g) Rigid body rotation; (h) Ripples; (i) Network pattern (this is an
example of splitting-merging behaviours in 2D)

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

As before, xi , vi ∈ Rn , x ∈ Rn and t ∈ R+ . The first term in Eq. (1.12b) is

a friction force (with ξ the friction coefficient of bacteria moving on the fixed
substrate). The second term in (1.12b) is a chemotactic attraction force due to the
chemical c, while the third term in (1.12b) is a random force (where Ri (t) is a
white noise and D is a diffusion coefficient). Finally, the terms in Eq. (1.12c)
describe the degradation of the chemical c at a rate k, the diffusion of this
chemical with diffusion coefficient Dc , and the production of the chemical at
a rate h by the cells themselves (i.e., at every spatial position of these cells).
Numerical simulations with this model have shown the formation of a network
pattern with filamentary structure; see also Fig. 1.2.
To conclude the discussion on these models, we note that the attractive effect
of the chemical could be replaced by an attractive-repulsive potential U [66]:

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

where m is the mass of particles/cells/bacteria. . . , ULR is a long-range binary

potential, USR a short-range binary potential, and Uext an external potential [66].
5. Zmurchok–de Vries Model. A final individual-based model that we review
here in the context of homogeneous populations was recently introduced by
Zmurchok and de Vries [67] to compare pattern formation with a 1D nonlocal
two-speed kinetic/hyperbolic models previously introduced in [68, 69] (which
will be discussed in detail in Chap. 5). The changes in the position (xi ∈ R)
and direction (vi ) of particles/cells/individuals are described by the following
equations in 1D:

xi (t + Δt) = xi (t) + Γi± (yi± )vi (t + Δt)Δt, i = 1, . . . , N, (1.15a)


−vi (t), if λ±i Δt ≥ X,
vi (t + Δt) = (1.15b)
vi (t), otherwise,

with xi ∈ R, t ∈ R+ , X a uniformly distributed random variable

on [0, 1] (which is updated at every time step), λ± i Δt the probability
of turning during the time step Δt, and Γi (yi± ) is the travel speed.
Here, λ±
i
±
= λ1 + λ2 f (yal,i ), Γi± (yi± ) = γ g(ya,i
± ±
− yr,i ±
), with yr,a,al;i
1.1 Modelling Self-organised Aggregation and Movement 11

describing the repulsive(r)/attractive(a)/alignment(al) social interactions for each

L,R L,R
particle/cell/individual i within the respective interaction zones Zr;i , Za;i
L,R
and Zal;i (see also Fig. 1.3). Note that the ± signs describe the direction of
the reference individual, while the superscripts L, R describe the interaction
zone to the Left or to the Right of this individual i. These social interaction
forces are defined the same way as the interaction forces for the hyperbolic
models in [68, 69]. For example, the attractive force experienced by the ith

left−moving individual at position x i right−moving individual at position x i

attraction, repulsion for "i" alignment for "i"
omnidirectional
and λ+i λ+i
unidirectional M1
reception of xi −si xi xi +s i xi −si xi x i+s i
stimuli

omnidirectional λ+i λ+i

reception
(stimuli from M2
all neighbors) xi −si xi xi +s i xi −si xi x i+s 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)

unidirectional λ+i λ+i

reception M4
(e.g., sound xi −si xi xi +s i xi −si xi x i+s i
signals)

unidirectional λ+i λ+i

reception M5
(e.g., tactile xi −si xi xi +s i xi −si xi x i+s i
signals)
ZLr ZRr ZLal ZRal

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

Fig. 1.3 Caricature description of five different directional 1D communication mechanisms

introduced in [69], which were adapted in [67] to describe the perception of neighbours by a
reference (left-moving/right-moving) individual positioned at xi . The turning rates λ±
i depend
on the perception of neighbours within the repulsion zone (ZrL,R ), alignment zone (Zal L,R
), or
L,R
attraction zone (Za )
12 1 Introduction

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

(a) (b) (c)

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
(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

2600 2600 2600

t

2400 2400 2400

2200 2200 2200

2000 2000 2000

0 5 10 0 5 10 0 5 10
x x x

Fig. 1.4 Examples of spatio-temporal patterns exhibited by the 1D individual-based mod-

els (1.15). Here N = 500 individuals that randomly scattered over a 1D domain [0, L], with
L = 10 and periodic boundary conditions. Patterns reprinted from [67], under the Creative
Commons Attribution License. (a) Stationary pulse 1. (b) Stationary pulse 2. (c) Ripples. (d)
Feathers. (e) Travelling pulse. (f) Travelling train. (g) Zigzag pulse. (h) Breathers. (i) Travelling
breather. Copyright c 2018 Zmurchok, de Vries

obtained in [69] with a hyperbolic model incorporating similar communication

mechanisms: from stationary pulses, to travelling pulses, travelling trains, ripples,
breathers, travelling breathers, feathers, zigzags, and splitting and merging pat-
terns. These patterns are summarised in Fig. 1.4. In Chap. 5 we will introduce a
kinetic/hyperbolic 1D model that is the mesoscale version of this microscale model,
and show how numerical simulations combined with bifurcation and symmetry
theory can help us understand the formation of some of the patterns, as well as the
transitions between certain patterns. Here, we emphasise that no investigation of the
phase transitions displayed by this individual-based model has been performed yet.
14 1 Introduction

Even if the previous individual-based models focused on homogeneous pop-

ulations, in nature the cell/bacterial/animal populations are not homogeneous.
Therefore, some recent studies have developed individual-based models that inves-
tigate the interactions between different populations [70–72]. In the following we
briefly discuss such an individual-based model, since the patterns generated by this
model (see also Fig. 1.5) are generic for the interactions between two populations

(a) population 1 (b) population 1

population 2 population 2
space (y)

(c) population 1 (d) population 1

population 2 population 2
space (y)

(e) (f)
space (y)

space (x) space (x)

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:

xi (t + Δt) = xi (t) + vi (t)Δt, i = iA + iB , iA,B = 1..NA,B , (1.17)

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

It was assumed in [72] that the concentration cT of a chemical signal is secreted

by two sources: a local chemical signal cI secreted by the cells themselves (at
every position xi of cells), and an externally imposed chemical gradient cE
(changing only along the y direction). Therefore, ∇cT = ∇cI + ∇cE , with


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

In the equation for the√gradient of cI , KI is a modified Bessel function of

the second kind, l = DI μI (with DI = diffusion rate of chemical cI , μI =
degradation rate of chemical cI ), p measures the response of the cells to the
chemical gradient, and xij is a unitary vector directed from j to i. In the equation
for the gradient of cE (obtained after solving a simple diffusion-reaction equation
for the degradation of a chemical), cm gives the constant concentration of the
chemical cE at the upper boundary of the domain (in the y-direction), DE is the
16 1 Introduction

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.

1.1.2 Kinetic and Macroscopic Models

The kinetic (mesoscale) models describe phenomena occurring at intermediate

scales, where one can investigate the behaviour of the whole population as it is
influenced by individual characteristics (e.g., individual’s velocity, age, size, etc.).
The macroscale models can be derived from the kinetic models by taking an average
over the characteristic variable. This way, the description of the statistics for the
characteristic variable (e.g., velocity) in the kinetic models is reduced in continuous
models to only a small number of its moments [78]. For both mesoscale and
macroscale models, the spatial scale of interest is at least one order of magnitude
larger than the diameter of a cell (or the size of an individual/particle/etc.).
Initially, the macroscale models for animal movement and aggregation were
mainly of parabolic type (see [73, 79–86], and the references therein). These
models are suitable when one is interested in measuring the population as a whole
(i.e., mean square displacements, mean population drift) [87]. However, since the
parabolic models unrealistically assume an infinite speed of propagation (i.e., even
if the initial data has compact support, perturbations in this data can be perceived
everywhere in the domain [88]), scientists started to focus more and more on
hyperbolic and kinetic models [69, 73, 87, 89–94]. The use of these hyperbolic
and kinetic models is helped by the recent advances in experimental techniques,
which allow scientists to track the movement of individual particles/cells/animals
(i.e., individual speeds and turning angle distribution) [95, 96]. When this kind of
information exists, the hyperbolic and kinetic models become the natural choice of
continuum models [87]. Another reason for using these two types of models is the
possibility of incorporating into variables detailed descriptions of various biological
functions and activities (e.g., different animal communication mechanisms [69],
or different degrees of cell differentiation or cell activation—information that is
characteristic to lower molecular scales [24]). Moreover, directed movement that
leads to alignment/polarisation is easier incorporated into these transport models.
In principle, the use of kinetic and hyperbolic models (as well as parabolic
models) requires having a large population of interacting organisms [44, 97, 98]. As
emphasised in [97], mesoscale/macroscale models for biological movement and pat-
tern formation usually incorporate the hidden assumption that statistical correlations
between cells/particles can be neglected, which means that the predictions of these
models might fail in those spatial regions where cell/particle numbers are small
and there are strong long-range correlations in cell/particle movement. This calls
for caution when applying such models to smaller populations of bacteria, cells, or
animals, since it could lead to different group dynamics (as observed by Chuang et
Another random document with
no related content on Scribd:
The Project Gutenberg eBook of The Cambridge natural
history, Vol. 02 (of 10)
This ebook is for the use of anyone anywhere in the United States and most
other parts of the world at no cost and with almost no restrictions whatsoever.
You may copy it, give it away or re-use it under the terms of the Project
Gutenberg License included with this ebook or online at www.gutenberg.org. If
you are not located in the United States, you will have to check the laws of the
country where you are located before using this eBook.

Title: The Cambridge natural history, Vol. 02 (of 10)

Author: Frank E. Beddard

W. B. Benham
F. W. Gamble
Marcus Hartog
Lilian Sheldon

Editor: S. F. Harmer
Sir A. E. Shipley

Release date: October 16, 2023 [eBook #71891]

Language: English

Original publication: New York: MacMillan and Co, 1901

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)

*** START OF THE PROJECT GUTENBERG EBOOK THE CAMBRIDGE

NATURAL HISTORY, VOL. 02 (OF 10) ***
 THE

CAMBRIDGE NATURAL HISTORY

EDITED BY

S. F. HARMER, M.A., Fellow of King's College, Cambridge; Superintendent of the

University Museum of Zoology

AND

A. E. SHIPLEY, M.A., Fellow of Christ's College, Cambridge; University Lecturer on

the Morphology of Invertebrates

VOLUME II

FLATWORMS AND MESOZOA

By F. W. Gamble, M.Sc. (Vict.), Owens College

NEMERTINES
By Miss L. Sheldon, Newnham College, Cambridge

THREAD-WORMS AND SAGITTA

By A. E. Shipley, M.A., Fellow of Christ's 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

EARTHWORMS AND LEECHES

By F. E. Beddard, M.A. (Oxon.), F.R.S., Prosector to the Zoological Society, London

GEPHYREA AND PHORONIS

By A. E. Shipley, M.A., Fellow of Christ's College, Cambridge

POLYZOA
By S. F. Harmer, M.A., Fellow of King's College, Cambridge

London
MACMILLAN AND CO., Limited
NEW YORK: THE MACMILLAN COMPANY
1901

All rights reserved

'Nous allons faire des vers ensemble'

André de Chénier

First Edition 1896. Reprinted 1901

 CONTENTS

PAGE
Scheme of the Classification adopted in this Book ix

PLATYHELMINTHES AND MESOZOA

CHAPTER I
TURBELLARIA
Introduction—description of the Polyclad Leptoplana
tremellaris—Appearance—Habits—Structure: Polycladida—
Classification—Habits—Anatomy—Development: Tricladida—
Occurrence—Structure—Classification: Rhabdocoelida—
Occurrence—Habits—Reproduction—Classification 3
CHAPTER II
TREMATODA
Characters of Trematodes—Habits and Structure of Trematoda
Ectoparasitica (Monogenea)—Life-Histories of Polystomum
integerrimum, Diplozoon paradoxum, and Gyrodactylus
elegans—Trematoda Endoparasitica (Digenea)—Occurrence
and Habits of Digenea—Life-History of Distomum
macrostomum—Distomum hepaticum and its Effects—Bilharzia
haematobia—Bisexual Trematodes—Table of Hosts—
Classification 51
CHAPTER III
CESTODA
Introduction—Nature of Cestodes—Occurrence of Cestodes—
The Tape-Worms of Man and Domestic Animals—Table of Life-
Histories of Principal Cestodes of Man and Domestic Animals
—Structure and Development of Cestodes—Table for the
Discrimination of the More Usual Cestodes of Man and
Domestic Animals—Classification 74
CHAPTER IV
MESOZOA
Dicyemidae—Structure—Reproduction—Occurrence:
Orthonectidae—Occurrence—Structure: Trichoplax:
Salinella 92

NEMERTINEA
 CHAPTER V
NEMERTINEA
Introductory—External Characters—Anatomy—Classification—
Development—Habits—Regeneration—Breeding—
Geographical Distribution—Land, Fresh-Water, and Parasitic
Forms—Affinities 99

NEMATHELMINTHES AND CHAETOGNATHA

CHAPTER VI
NEMATHELMINTHES
Introduction—Nematoda—Anatomy—Embryology—Classification
—Ascaridae—Strongylidae—Trichotrachelidae—Filariidae—
Mermithidae—Anguillulidae—Enoplidae—Parasitism:
Nematomorpha—Anatomy—Classification—Life-History:
Acanthocephala—Anatomy—Embryology—Classification 123
CHAPTER VII
CHAETOGNATHA
Structure—Reproduction—Habits—Food—Classification—Table
of Identification [see also p. 534] 186

ROTIFERA, GASTROTRICHA, AND KINORHYNCHA

CHAPTER VIII
ROTIFERA, GASTROTRICHA, AND KINORHYNCHA
Rotifera—History—External Features—Movements—Anatomy—
Reproduction—Embryology—Classification—Distribution—
Affinities: Gastrotricha: Kinorhyncha 197

ARCHIANNELIDA, POLYCHAETA, AND MYZOSTOMARIA

CHAPTER IX
The Chaetopodous Worms—The Archiannelida—Anatomy of
Nereis, as Typical of the Polychaeta 241
CHAPTER X
Classification of the Polychaeta—Shape—Head—Parapodia—
Chaetae—Gills—Internal Organs—Jaws—Sense Organs—
Reproduction—Larval Forms—Budding—Fission—Branching—
Regeneration 257
CHAPTER XI
Natural History of Polychaetes—General Habits—Character of 284
 Tube and its Formation—Colouring—Protective and Mimetic
Devices—Phosphorescence—Food—Uses—Associated Worms
—Worms as Hosts—Distribution—Fossil Remains
CHAPTER XII
Characters of the Sub-Orders of Polychaetes—Characters of
the Families—Description of British Genera and Species: the
Myzostomaria 303

OLIGOCHAETA (EARTHWORMS, ETC.), AND HIRUDINEA

(LEECHES)
CHAPTER XIII
OLIGOCHAETA (EARTHWORMS AND THEIR ALLIES)
Introduction—Anatomy—Reproduction—Bionomics—Distribution
—Classification—Microdrili and Megadrili 347
CHAPTER XIV
HIRUDINEA (LEECHES)
Introduction—Anatomy—Reproduction—Classification—
Rhynchobdellae and Gnathobdellae 392

GEPHYREA AND PHORONIS

CHAPTER XV
GEPHYREA
Introduction—Anatomy—Development—Sipunculoidea—
Priapuloidea—Echiuroidea—Epithetosomatoidea—Affinities of
the Group 411
CHAPTER XVI
PHORONIS
History—Habits—Structure—Reproduction—Larva—
Metamorphosis—List of Species and Localities—Systematic
Position 450

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).

NEMATHELMINTHES (p. 123)

Ascaridae (p. 138).
Strongylidae
(p. 142).
Trichotrachelidae
(p. 144).
Filariidae (p. 147).
Mermithidae
NEMATODA (pp. 123, 124) (p. 150).
Anguillulidae
(p. 154).
Enoplidae (p. 157).
Chaetosomatidae
(p. 158).
Desmoscolecidae
(p. 159).

NEMATOMORPHA (pp. 123, 164) Gordiidae (p. 164).

Echinorhynchidae
(p. 182)
Gigantorhynchidae
(p. 183).
ACANTHOCEPHALA (pp. 123, 174)
Neorhynchidae
(p. 184).
Arhynchidae
(p. 185).

CHAETOGNATHA (p. 186)

ROTIFERA (p. 197)

FLOSCULARIACEAE (p. 220) Flosculariidae
(p. 221).
Apsilidae (p. 221).

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)

KINORHYNCHA (p. 236)

CHAETOPODA (p. 241)

ARCHIANNELIDA (p. 241)

POLYCHAETA Phanerocephala Syllidae (p. 306).

(pp. 241, 245) (p. 303) Hesionidae (p. 308).
Aphroditidae
(p. 309).
Phyllodocidae
(p. 313).
Tomopteridae
(p. 315).
Nereidae (p. 315).
Nereidiformia Nephthydidae
(p. 303) (p. 317).
Amphinomidae
(p. 318).
Eunicidae (p. 318).
Glyceridae (p. 320).
Sphaerodoridae
(p. 320).
Ariciidae (p. 321).
Typhloscolecidae
(p. 321).
Spionidae (p. 321).
Polydoridae
(p. 323).
Spioniformia Chaetopteridae
(p. 304) (p. 323).
Magelonidae (325.
Ammocharidae
(p. 325).
 Terebelliformia Cirratulidae (p. 325).
(p. 304) Terebellidae
(p. 327).
Ampharetidae
(p. 330).
Amphictenidae
(p. 330).
Capitelliformia Capitellidae
(p. 305) (p. 331).
Opheliidae (p. 331).
Maldanidae
(p. 332).
Arenicolidae
(p. 333).
Scoleciformia
Scalibregmidae
(p. 305)
(p. 334).
Chlorhaemidae
(p. 334).
Sternaspidae
(p. 335).
Sabellidae (p. 336).
Eriographidae
Sabelliformia (p. 338).
Cryptocephala (p. 305) Amphicorinidae
(p. 303) (p. 339).
Serpulidae (p. 339).
Hermelliformia Hermellidae
(p. 306) (p. 341).

MYZOSTOMARIA (pp. 241, 341)

OLIGOCHAETA Microdrili (p. 373) Aphaneura (p. 374).

(pp. 241, 347) Enchytraeidae
(p. 375).
Discodrilidae
(p. 376).
Phreoryctidae
(p. 376).
Naidomorpha
(p. 377).
Tubificidae (p. 378).
Lumbriculidae
(p. 379).
 Moniligastridae
(p. 380).
Perichaetidae
(p. 380).
Cryptodrilidae
(p. 382).
Acanthodrilidae
Megadrili (pp. 373, 374). (p. 384).
Eudrilidae (p. 385).
Geoscolicidae
(p. 386).
Lumbricidae
(p. 388).

HIRUDINEA (p. 392)

Ichthyobdellidae
(p. 406).
RHYNCHOBDELLAE (p. 405)
Glossiphoniidae
(p. 406).

Gnathobdellidae
(p. 407).
GNATHOBDELLAE (p. 407)
Herpobdellidae
(p. 407).

GEPHYREA (p. 411)

SIPUNCULOIDEA (pp. 412, 420).
PRIAPULOIDEA (pp. 412, 430).
ECHIUROIDEA (pp. 412, 434).
EPITHETOSOMATOIDEA (pp. 412, 444).

PHORONIS (p. 450)

POLYZOA (p. 465)

ENTOPROCTA (pp. 475, 487)

ECTOPROCTA Gymnolaemata Cyclostomata Articulata (p. 517).

(p. 475) (p. 476) (p. 477) Inarticulata (p. 517).
 Cheilostomata Cellularina (p. 518).
(p. 477) Flustrina (p. 518).
Escharina (p. 518).
Alcyonellea (p. 518).
Ctenostomata
Vesicularina
(p. 477)
(p. 518).
Phylactolaemata (pp. 476, 493)
 PLATYHELMINTHES AND MESOZOA

BY

F. W. GAMBLE, M.Sc. (Vict.)

Demonstrator and Assistant-Lecturer in Zoology in the Owens College, Manchester.

CHAPTER I

TURBELLARIA

INTRODUCTION: DESCRIPTION OF THE POLYCLAD LEPTOPLANA TREMELLARIS—

APPEARANCE—HABITS—STRUCTURE: POLYCLADIDA—CLASSIFICATION—HABITS—
ANATOMY—DEVELOPMENT: TRICLADIDA—OCCURRENCE—STRUCTURE—
CLASSIFICATION: RHABDOCOELIDA—OCCURRENCE—HABITS—REPRODUCTION—
CLASSIFICATION.

The Platyhelminthes, or Flat Worms, form a natural assemblage of animals, the

members of which, however widely they may differ in appearance, habits, or life-
history, exhibit a fundamental similarity of organisation which justifies their
separation from other classes of worms, and their union into a distinct phylum.
Excluding the leeches (Hirudinea), and the long sea-worms (Nemertinea)—which,
though formerly included, are now treated independently—the Platyhelminthes
may be divided into three branches: (1) Turbellaria (including the Planarians), (2)
Trematoda (including the liver-flukes), and (3) Cestoda (tape-worms). The
Mesozoa will be treated as an appendix to the Platyhelminthes.

The Turbellaria were so called by Ehrenberg[1] (1831) on account of the cilia or

vibratile processes with which these aquatic animals are covered, causing by their
incessant action, tiny currents ("turbellae," disturbances) in the surrounding water.
The ciliary covering distinguishes this free-living group from the parasitic
Trematodes and Cestodes, some of which possess such an investment, but only
during their early free larval stage, for the short period when they have left the
parental host and are seeking another (Figs. 26, 27, 42).

Some Turbellaria (Rhabdocoelida) resemble Infusoria in their minute size, shape,

and movements. Nevertheless they possess an organisation of considerable
complexity. The fresh-water Planarians (Fig. 14), abounding in ponds and streams,
vary from a quarter to half an inch in length, and are elongated and flattened. Their
body is soft, and progresses by a characteristic, even, gliding motion like a snail.
The marine Planarians or Polyclads (Fig. 8) are usually broad and leaf-like,
sometimes attaining a length of six inches, and swim or creep in a most graceful
way. Land Planarians occur in this country (Fig. 15), but far more abundantly in
tropical and sub-tropical districts, in moist places, venturing abroad at night in
pursuit of prey. They are elongated and cylindrical, in some cases measuring,
when fully extended, a foot or more in length, and are often ornamented with
brilliantly coloured, longitudinal bands.

Turbellaria are carnivorous, overpowering their prey by peculiar cutaneous

offensive weapons, and then sucking out the contents of the victim by the
"pharynx." Land Planarians feed on earthworms, molluscs, and wood-lice; fresh-
water Planarians on Oligochaet worms, water-snails, and water-beetles; marine
forms devour Polychaet worms and molluscs. Some Turbellaria seem to prefer
freshly-killed or weakly examples of animals too large to be overpowered when
fully active. Certain Rhabdocoelida are messmates of Molluscs and Echinoderms,
and a few others are truly parasitic—a mode of life adopted by all Trematodes save
Temnocephala.

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.

The Cestodes or Tape-worms have undergone more profound modifications both in

structure and in mode of development. They are all endoparasitic, and, with one
exception (Archigetes), attain maturity solely within the alimentary canal of
Vertebrates. In length they range from a few millimetres to several metres, but this
great size is attained from the need for the rapid production and accumulation of
enormous numbers of eggs. The "head" or "scolex" is attached to the mucous
membrane of the host by suckers or hooks, but there is no mouth nor any certain