Complex Dynamics and Interaction Networks

in Animal Societies (DYNACTOM)

 

Main research axes

 

• Physiological bases of social behaviours

• Coordination mechanisms and dynamics of collective motion in animal groups

• Bio-inspired models of swarm intelligence

• Task allocation mechanisms and division of labour in ants

• Morphogenesis and collective building in ants and termites

 

 

 

Physiological bases of social behaviors Richard Bon, Audrey Dussutour , Raphaël Jeanson

 

In this project, we study the physiological determinants involved in the emergence of sociality and regulation of collective behaviour in different organisms at different time scales. This proposal is integrative, spanning neuropharmacology to behaviour in an evolutionary frame.

 

On the short-term scale, we study the relation between social context and the physiological expression of stress in ungulates. The effect of group size at the collective level, and on the activity budget and basal level of stress is measured. We determine if increasing group size induces a social buffering effect in stressing situations. We then investigate whether a difference in stress expression exists between individuals and if it affects the collective organisation. Finally, we experimentally manipulate the stress level at an individual level and observe the outcome at the collective level.

Food distribution inside an ant colony by way of an interaction network   Green-headed ants feeding on a droplet of sugar   Behavioural observation of a group of grazing sheep

On the mid term scale, we determine the role of the individual nutritional state on social cohesion, division of labor and colony growth in ants. Ants and all social insects are faced with an additional nutritional challenge compared to solitary species: the food entering a social insect colony is brought by only a small number of its workers and is shared among all members of the colony. We study how ants adjust their harvesting strategy to meet the total demand for nutrients within the nest. We then investigate how nutritional needs vary between workers according to both their cast and their task. Finally, we examine the role of food composition in cast determinism and colony growth.     On the long-term scale, we examine the fundamental mechanisms governing the emergence of cooperation amongst individuals and driving the formation of perennial societies in spiders. We manipulate the environmental constraints in solitary species to obtain artificial societies and to identify the physiological determinants underlying the expression of phenotypic plasticity leading to the formation, maintenance and inheritance of these social structures. First we investigate how food availability affects physiological state and social tendency in spiderlings. Second we determine the contribution of maternal effect in various nutritional environments on the spiderlings' social development.

 

References

• Dussutour A. and Simpson S. J  2009 " Communal nutrition in ants" Current Biology  19,740-744

Top

 

 

Coordination mechanisms and dynamics of collective motion in animal groups Richard Bon, Maud Combe, Audrey Dussutour, Vincent Fourcassié , Jacques Gautrais, Mathieu Moreau, Mehdi Moussaïd, Marie-Hélène Pillot, Guy Theraulaz

 

Many animals move in groups of large size that can reach hundred to thousand individuals. How do individuals interact within these groups so as to ensure cohesive movements ? What are the behavioural mechanisms involved in the coordination and dynamics of their collective movements ? The research carried out in our team addresses these questions for different types of collective movement and in different species, invertebrates as well as vertebrates.

 

Prior to the establishment of a recruitment trail ants have to search for and find food. As a first step to the understanding of the collective search behaviour of interacting ants, we have begun to study the food searching behaviour of individual workers of various ant species to see how it fits with the different theoretical models of search strategies currently proposed in statistical physics.

 

Ants use simple behavioural rules to organize and regulate their traffic on recruitment trails (Dussutour et al 2009, Fig 1). We plan to explore these rules and to study in particular the possible role of physical contacts between ants (Dussutour et al 2007). We also plan to investigate whether ants are able to collectively solve complex geometrical problems through their trail-recruitment process, e.g. whether they are able to establish their trail along the path of lesser effort.

Fig 1: The collective choice of a path in the ant Lasius niger When given the choice between two equal paths leading to the same goal ants collectively choose a single path

Fig 2: Self-organized formation in a pedestrian flow The flow of pedestrians in a crowded street spontaneously self-organizes in lanes of individuals moving in opposite direction

The motion of human crowd results from self-organized processes based on local interactions among individuals (Mousaïd et al 2009, Fig 2). We investigate the collective motion of pedestrians through computer simulations based on models in which the movement of individuals obeys simple behavioural rules of avoidance. The validity of these rules is then tested experimentally. In an another set of experiments we plan to investigate the conditions under which lane segregation (spatial separation between opposite flows) occurs in pedestrian flows.

 

Some animals such as fish or sheep can form groups comprised of up to several thousands individuals moving and interacting together (Fig 3 & 4). We use specially designed experimental set-ups to track the trajectories of individual fish or sheep moving in groups of various size (2 to 100 individuals) and to quantify their interactions (Gautrais et al 2007, 2009; Michelena et al 2006). These data will be used to formulate incrementally a model of the key elements in the stimulus / response function of the individuals. We plan also to study whether some individuals can be more influential than others in driving the movement of the group.

 

Fig 3: Self-organized formation in a fish shoal
(a) Emergence of a vortex in a school of barracudas, consisting of individuals circling around an

unoccupied core. (b) Some simple rules used by individuals to synchronize with their neighbours

can lead to emergent spatio-temporal patterns at large scale

 

 

Fig 4: Self-organized formation in a group of sheep

 

 

References

• Dussutour, A., Beshers, S., Deneubourg, J.L. & Fourcassié V.. Crowding increases foraging efficiency in the leaf-cutting ant Atta colombica. Insectes Sociaux, 54: 168-165 (2007).

Top

 

 

Bio-inspired models of swarm intelligence Jacques Gautrais, Guy Theraulaz

 

In swarm robotics, very few works have been done to develop distributed algorithms for the coordination of motion of groups of robots. Most of them were based on the simulation of theoretical models; their implementation on real robots cannot be done without solving important algorithmic and communication problems. A new behavioural algorithm based on a recent model of collective motion in fish school will be implemented in a group of micro-robots.

 

This robotic setup will then be used to investigate the influence of noise and several physical parameters on the resulting collective behaviours, especially the transition between swarming and schooling. Besides providing a simple control algorithm for the coordinated motion of groups of robots in the context of biological robotics, we expect to test and validate some of the hypotheses on information processing at the individual level in the new models of fish school.

References

• Bonabeau, E., Dorigo, M., & Theraulaz, G. 1999. Swarm intelligence: From natural to artificial systems. Oxford University Press, Oxford, NY, USA.

Top

 

 

Division of labour and caste ergonomics in ants Patrick Arrufat, Audrey Dussutour,Vincent Fourcassié, Raphaël Jeanson , Jean-Paul Lachaud, Mathieu Moreau

 

Division of labour is a fundamental property of eusociality. It is based on the idiosyncrasy of group members and is coupled with age polyethism and sometimes with the existence of size polymorphism. We study the influence of colony size and social networks dynamics on task allocation and division of labour in societies of various ant species. We also consider the influence of the environment on the efficiency of division of labour (i.e. caste ergonomics) within these societies.

 

The efficiency of offensive weapons shapes hunting behaviour in most predators. Studies of hunting strategies in ants suggest that risk-prone predatory behaviour could be related to colony size. We test the influence of this variable on the choice of hunting strategy (mandible strike or stinging) by comparing colonies of the ant species Odontomachus opaciventris at different developmental stages and other Odontomachus species which occur naturally with contrasted differences in colony size.

The organisation of animal societies depends on the rate of interactions and information transfer among group members. In the ant O. hastatus, we examine the link between division of labour within colonies and the properties of the social network formed by the relationships among individuals. Using RFID technology to tag individually all workers in experimental colonies we test  how variations in group size, level of starvation, and the presence/absence of the queen influence task allocation and the properties of social networks.

a) Worker of Odontomachus opaciventris displaying a hunting posture. b) Worker of O. hastatus equipped with a RFID transponder (length=4 mm). c) Network of social interactions in a colony of 55 workers + queen of O. hastatus

 

 

A high degree of specialization in functional tasks, with the presence of elite specialists, is generally the rule in ant societies. However, the species Gigantiops destructor lacks such high specialization for "prey chewing". This suggests that this task could be important to maintain the social cohesion of the colony. We address this issue by studying the impact of this behaviour on the social organization and the structure and properties of interaction networks within colonies (presence/absence of key individuals).

 

In Messor barbarus transport efficiency decreases more rapidly for increasing load size

on a rough substrate compared to a smooth substrate in media than in minor workers.

 

 

The ergonomic efficiency of insect societies not only depends on their intrinsic characteristics, such as their size and composition, but also on the constraints of the environment. Working with ant species transporting external loads such as seeds or leaf fragments, we examine how the interplay between polymorphism and the structural complexity of the environment can affect the efficiency of transport in ants, both at the individual and colony level.

 

Lack of elite specialists for "prey chewing" behaviour in a colony of G. destructor

 

Top

 

 

Morphogenesis and Collective Construction Behaviour Maud Combe, Jacques Gautrais, Christian Jost , Anaïs Khuong, Chaker Sbaï, Guy Theraulaz

 

Nests collectively built by social insects have often very complex designs (Fig 1) and are ingeniously adapted to the colony's needs. A growing literature tries to understand the emergence of such complex structures through self-organized individual behaviours coupled to environmental information. Based on our work on clustering behaviour (Theraulaz et al 2002) and its modulation by the environment (Challet et al 2005, Jost et al 2007, Casellas et al 2008) we ask now how social insect nests emerge from the individual construction behaviours.

 

Fig 1: Example nest architectures and their virtual reconstructions

(a) Apicotermes sp. (subterranean), (b) Sphaerotermes sphaerothorax (subterranean wall fragment)

 

Termites and ants build a variety of architectures. We are building a data base of X-ray tomographies of such architectures and develop (in collaboration with physics, informatics and mathematics laboratories) the tools to describe and characterize such nests.

 

Fig 2:Characterization of nest architectures (a) Cubitermes nest with its virtual reconstruction from X-ray tomography data, (b) and its description as a communication network (each chamber is a node, each tunnel an edge).

Cubitermes type structures (Fig 2) were already successfully described as communication networks (Perna et al 2008a,b). Other types of architectures, such as the sponge-like Nasutitermes or Trinervitermes nests, will require other well adapted tools.   How do these architectures emerge from the individual construction behaviours? We will extract the individual statistics (e.g. the probability to pick up and to drop a pellet) and express these statistics as a function of local environmental parameters.

We will use the native ant Lasius niger as a model animal to address these questions (Fig 4). L. niger nests consist both of hypogeous excavated parts and epigeous constructed parts. We want to understand the early stages of the construction behaviour where we see the emergence of pillars and roofs. These structures will be characterized by surface scanner technology (Fig 4) and compared to the simulated predictions from the minutely modelled individual behaviour (Fig 3).

Fig 4: Construction dynamics and characterization of spatial patterns (a) Quantification of Lasius niger constructions in the laboratory with a surface scanner, (b) Result after 2 weeks with regularly spaced pillars constructed on a plaster support.

 

Displacement in these nests is of primordial importance to the colony’s survival (food transport, nest defence, brood care in the proper micro climates). Real displacement along the nest’s walls may be rather different from idealized displacement in the network description. Borders were already shown to play a fundamental role in the animal’s distribution in space (Casellas et al 2008). Mathematical and simulation modelling of such displacement (Fig 3 shows snapshots of an individual based simulation platform) will further our understanding of nest function.

 

Fig 3: Individual based simulation of ant construction
(a) Constructions of Lasius niger observed under laboratory conditions,

(b-c) construction in the simulator platform at two different stages of construction

(red "pellets" are the head positions of moving ants).

 

 

The know-how acquired with the Lasius experimental paradigm will also be put to work with termite construction behaviour. A collaboration with two laboratories in Brazil has already started in order to study the nests and the construction behaviours of the sympatric Cornitermes cumulans and Procornitermes araujoi (Fig 5).

 

Fig 5: Cornitermes cumulans nests (Cerrado, Brazil)
(a) Workers at construction work, (b) A partially excavated nest, (c) A smaller nest cut in half.

 

 

References

• Casellas E., Gautrais J., Fournier R., Blanco S., Combe M., Fourcassié V., Theraulaz G. & Jost C. From individual to collective displacements in heterogeneous environments. Journal of Theoretical Biology, 250, 424-434 (2008). doi: 10.1016/j.jtbi.2007.10.011