APS 323 Social Insects: Lecture 12

1
APS 323 Social Insects Lecture 12
Francis L. W. Ratnieks Laboratory of Apiculture
Social Insects
Department of Animal Plant Sciences University
of Sheffield
Lecture 12 Principles of Self Organization
2
Aims Objectives
Aims 1. To explain the general principles behind
self-organization. 2. To give specific examples
in the balancing of foragers and receivers in
honey bee colonies and in the pheromone trails of
ants. Objectives 1. Learn and understand the
basic principles 2. Learn the examples in
relation to the principles
3
How Do Insect Societies Organize the Many Workers?
4
Army Ant Colony
5
Maurice Maeterlinck
Maurice Maeterlinck (1862-1949) Nobel Prize in
Literature 1911 La Vie des Abeilles The Life of
the Bee 1900 What is it that governs here? What
is it that issues orders, foresees the future,
elaborates plans, and preserves equilibrium?
6
Maurice Maeterlinck
His book is not an abstract of natural history
but an exuberantly poetic work abounding in
reflections, the sum total of which is almost a
declaration of incompetence. It is useless, the
author seems to say, to inquire if the strange
cooperation among the bees, their apportionment
of work, and their social life are the product of
a reasoning mind. It matters little whether the
term instinct or the term intelligence is
used, for they are but ways of revealing our
ignorance in the matter. What we call instinct
among the bees is perhaps of a cosmic nature, the
emanation of a universal soul. One immediately
thinks of Virgil's immortal description of the
bees in which he says that a thinker attributes
to them a share of divina mens, the divine
thought, the divine spirit. From citation by
Nobel Committee
7
Bible BT
Go to the ant, thou sluggard consider her ways,
and be wise. Which having no guide, overseer of
ruler, provideth her meat in the summer, and
gathereth her food in the harvest. The Bible,
Proverbs 6 6-8
8
Self-Organization Basic Concepts
9
Camazine, S., Deneubourg, J.-L., Franks, N. R.,
Sneyd, J., Theraulaz, G., Bonabeau, E. 2001.
Self-organization in biological systems.
Princeton University Press.
10
Self Organisation
System composed of many sub-units or
agents workers in an ant colony cells in a
multicellular organism grains of sand on a
beach Individual agents react to local
conditions Global pattern arises from agent
behaviour No one is in overall charge No one
knows the overall state of the system
11
Other Ways of Organizing
Imagine that a group of individuals are
constructing a nest or structure, or are trying
to organize themselves in some particular way.
There are several possible mechanisms that do not
use self organization. Leader (central
control) Follow instructions of a
leader Blueprints e.g., use of a plan to
construct a building Recipe Follow a set of
instructions Templates e.g., use of a mould or
full-size guide Camazine et al. 2001
12
Mexican Wave
A Mexican wave is a good example of a global
pattern that can be generated either by self
organization (i.e., lift you arms when the person
to one side of you does), or by central
control/having a leader (i.e., lift your arms
when instructed to do so by the leader).
13
Mexican Wave
14
Advantages of Self-Organization
Simple Simple to implement. The global pattern
can arise from relatively simple rules by which
the agents interact with each other and their
environment. Robust The system is robust to
losing or gaining agents. Possible Central
control may not be easy to achieve in an insect
society. How would a worker or group of workers
who were in charge gather information from other
workers on what was going on in the colony,
process the information, and communicate back to
the other workers telling them what to do?
15
Pheromone Trails to Food in Ants
16
Army Ant Colony
17
Pharaohs Ant Trails
18
Self Organisation of Ant Trails
Goal Trails Lead to food How could this come
about via self organization? Individual foragers
react to local conditions foragers are walking
around outside nest if a forager finds food,
she lays trail back to the nest if a forager
finds a trail, she follows it to the food
Global pattern arises from agent behaviour
trails leading to food sources are established
this is an adaptive pattern No ant is in overall
charge No one ant knows the overall state of
the system
19
Positive Negative Feedback
There is both positive and negative feedback in
the trail system, meaning that more or less ants
can be directed to particular locations. Positive
feedback attract more ants to rewarding
locations Successful foragers lay pheromone
Trails attract more foragers Negative feedback
do not attract ants when food is used up
Unsuccessful foragers do not lay pheromone
Existing pheromone evaporates
20
Fungal Mycelium
21
Cellular Automaton Forest Fire Simulation
StarLogo Media Lab, MIT www.media.mit.edu/starlo
go
22
Cellular Automaton Ant Foraging Simulation
StarLogo Media Lab, MIT www.media.mit.edu/starlo
go
23
Cellular Automaton
This is a type of simulation method in which
space is divided up into many cells. The
properties of these cells are updated over and
over again. In the forest fire, a space can
either be empty (black) or containing a tree
(green). A tree on fire (red) can spread the fire
to a neighbouring cell if it contains a tree.
However the fire cannot spread into an empty cell
or a cell that has already burned (brown). The
ant foraging simulation follows the rules
outlined previously.
24
Natural Selection
Natural selection will favour genes that result
in ant colonies that collect food more
efficiently, as these colonies will survive and
reproduce better. Natural selection does not
select genes that directly result in global
properties of a trail system. No such genes
exist. Instead, natural selection can select
genes that affect the behaviour of workers, their
local rules and responses, which in turn
influence the global properties of the trail
system. Tendency of workers to lay pheromone
Amount of pheromone deposited Tendency of
workers to follow an existing trail etc.
25
Real Eciton Army Ant Raid Patterns
E. hamatum
E. rapax
E. burchelli
Different species of Eciton army ants have
different raid patterns. On the right is a broad
raid pattern suited for capturing poorly-defended
prey in a wide area. On the left a pattern that
concentrates ants in small areas, which is suited
for attacking well-defended prey such as nests of
ants and termites. In the centre is an
intermediate pattern.
26
Simulated Army Ant Raid Patterns
Different raid patterns that look very similar to
those of Eciton army ants can be generated in a
computer simulation by adjusting the local rules
used by workers. Deneubourg, J.-L., Goss, N.,
Franks, N., Pasteels, J. M. 1989. The blind
leading the blind modelling chemically mediated
army ant raid patterns. Journal of Insect
Behaviour. 3719-725.
27
Sand Dune
Self-organization is not confined to biological
systems. Many physical systems involve self
organization. The sand grains in a sand dune form
distinctive patterns according to the way that
they interact with each other and their
environment. However, natural selection has
not been altering the behaviour of sand grains to
create more adaptive sand patterns.
28
Balancing Nectar Foragers Receivers in Honey
Bee Colonies
29
Nectar Transfer in Hive
receiver
forager
30
Task Partitioning
Division of Labour Workers / Tasks Task
Partitioning Task / Workers
Ratnieks Anderson,1999, Insectes Sociaux
31
Balancing Work Capacities
Too Few Effect Feedback Signal Foragers
Short transfer delays Waggle dance Receivers
Long Transfer delays Tremble Dance
32
Where Tremble Dance is Made
Waggle dances are mainly made in the dance floor
area near the entrance. Tremble dances are made
further into the nest, where there are younger
bees. (Seeley 1995)
33
Effect of Delay on Forager
A returning nectar forager is more likely to make
the tremble dance, which recruits more nectar
receivers, if she has had a long delay (search
time) in being served by a receiver bee. (Seeley
1995)
34
Balancing Foragers Receivers
In honey bee colonies, nectar foragers do not
unload their nectar directly into cells. Instead,
they transfer it to receiver ( storer) bees who
then place it in cells. This is an example of
task partitioning. That is, the task of
collecting and storing each load of nectar is
divided into collecting and storing
sub-tasks. For the system to function efficiently
there must be a balance in the numbers of
foragers and receivers.This is achieved through
self-organization, based on the reactions of
foragers to the length of time they wait to be
unloaded. If a forager waits a long time it is
more likely to make the tremble dance. This
recruits in-nest bees to act as receivers. If the
delay is short a forager is more likely to make
the waggle dance, which recruits additional
workers to foraging and tells them where the
flowers are. Foragers often transfer their nectar
to several receivers. They probably do this to
gain a better estimate of the average delay in
being served. Each forager can only base its
decision about whether to make a waggle dance,
tremble dance, or no dance on its own local
experience and it can also be lucky or unlucky in
how rapidly it is served. Hence the value of
unloading to several receivers.
35
Phase Transition in Foraging Trail Organization
in Pharaohs Ants
36
Ant Density Trail Formation
Successful returning forager lays evaporating
trail
A few ants cannot lay a continuous trail because
of evaporation
Many ants can lay a continuous trail despite
evaporation
37
Phase Transition in Ferromagnet
ordered state
disordered state
38
Modelling Pheromone Trail Use
Use differential equations to determine the
number of foragers using the trail ASSUMPTIONS Pr
obability of forager JOINING trail increasing
function of trail strength Probability of forager
LEAVING trail decreasing function of trail
strength RESULTS Find the equilibria
where Number JOINING Number LEAVING
Beekman, Sumpter Ratnieks, 2001, PNAS
39
Equilibrium Trail Use
Type 2 Phase Transition (1 equilibrium no
hysteresis)
Number of foragers using trail
Type 1 Phase Transition (3 equilibria hysteresis)
Number of foragers
Beekman, Sumpter Ratnieks, 2001, PNAS
40
Stable Unstable Equilibria
1 Equilibrium 1 stable
3 Equilibria 1 unstable 2 stable
41
Pharaoh's Ant Monomorium pharaonis
syrup feeder
trail to feeder
nest box
42
Colony Size Increased Foraging
Increase in number of ants on trail after placing
syrup feeder
Beekman, Sumpter Ratnieks, 2001, PNAS
Colony size, ants
43
Helping a Trail to Get Started
final position of feeder
initial position of feeder
44
Testing for Hysteresis
Colony of 300 ants
Unhelped Helped
Colony of 700 ants
Beekman, Sumpter Ratnieks, 2001, PNAS
45
Phase Transition in Trail Formation
number of ants on trail
organised state adaptive
disorganised state non-adaptive
number of ants
46
Phase Transition in Pharaohs Ants
We discussed positive and negative feedback in
ant trails and self organization. In Pharaohs
ants, Monomorium pharaonis, the positive feedback
of more ants laying pheromone to food results in
a successful trail developing only if there is a
sufficient number of ants. Too few ants cannot
lay a trail given that pheromone is constantly
evaporating. Mathematical modelling shows that a
phase transition is expected. The model is based
on two simple and valid assumptions. An ant is
less likely to leave and more likely to join a
stronger trail. The model uses differential
equations to determine the equilibrium number of
foragers that follow the trail versus foraging on
their own. Experiment supports this prediction.
Colonies below 700 ants cannot recruit additional
foragers to a new food source. The model also
shows that at some colony sizes there may be two
stable equilibria (and one unstable equilibrium).
The two stable equilibria correspond to a colony
that has too few ants to start a trail but can
sustain a trail if one has already been started.
An experiment in which a colony is helped to form
a trail supports this prediction. Beekman, M.,
Sumpter D. J., Ratnieks F. L. W. 2001. Phase
transition between ordered and disordered
foraging in Pharaohs ants. Proceedings of the
National Academy of Sciences of the USA
989703-9706.