Navigation  without Java Scripts

STUDY OF ROUTE SELECTION DYNAMICS AND LOCAL MIGRATION OF AVI-FAUNA WITHIN A REAL ECOSYSTEM USING LEM MULTI-AGENT SIMULATOR

Anju Dahiya* , Serguei Krivov** and Jaweed Ashraf *
$ * : Lab. 317, School of Life Sciences,
Jawaharlal Nehru University,
New Delhi-110067 INDIA e-mail: a_dahiya@hotmail.com
**: scholar of Intercultural Open University, The Netherlands
present add: Indian Inst. of Ecology and Environment
A-15, Paryavaran Complex,
New Delhi 110030 INDIA e-mail: s_krivov@hotmail.com
$ : address for correspondence

 Abstract

We present a multiagent model using LEM simulator that enables one to explore the influence of individual behavior on the dynamics and structural complexity of route selection and local migration of animals within a real ecosystem over real time. The individuals act according to simple, biologically plausible rules. Here we present the results of a series of simulation experiments on space utilization by peafowl that is also used to calibrate the simulator against real-world systems and to demonstrate the simulator's strength for ecological modeling. This work involved a qualitative analysis of local routes selected by the peafowl within the limited space. We also explored the consequences of allowing individuals to make active choices about space selection by assigning weights and found that individuals had to adapt realistic behavioral ecological strategies. Our results lead us to believe that a better understanding of space utilization in real-world systems will require new approaches that use all aspects of Ecology i.e. the animal behavioral ecology along with population and community ecology along with mathematical treatment of these aspects.

Key Words: route selection dynamics; individual-based behavior; multi-agent ecological simulator.

1. Introduction

In our research work, we are trying the use of multi-agent simulation to study complex phenomena emerging in real ecological systems. The problem of analysis of ecosystem is a multifold one. It requires considerable efforts in experimental study of the ecosystem as well as theoretical work in the fields of ethology, mathematics, and computer science. Earlier we elaborated the main ethological principles in the context of computer modeling of adaptive behavior (Krivov et al. [5]). In this paper we present a multiagent simulation model of local migration of avi-fauna (Indian peafowl selected out of the survey of a number of birds) within the natural ecosystem of Jawharlal Nehru University (JNU), New Delhi Campus.

The dynamics of local migration of avi-fauna could be studied by creating an artificial universe of robotic birds and animals via computer simulation and conducting experiments in the so created universe. This approach is termed as ‘multiagent system simulation’ or as ‘individual based modeling’ and has emerged only during the last decade (see: Hraber et. al. [12]; Drogoul and Ferber [2]). A multiagent system typically consists of many dispersed agents acting in parallel with no global controller responsible for the behavior of all agents. Rather, the actions of each agent depend upon the states and actions of a limited number of other agents, and competition and coordination among the agents is subject to structural constraints. This complex determines the overall direction of the system. The behavior of the whole system tends to arise more from the interactions among the agents than from any constraints imposed into structure of the whole system as such. Moreover, the local interaction networks connecting the agents are continuously recombined and revised. In particular, niches that can be exploited by particular adaptations are continuously created, and their exploitation in turn leads to new niche creations, so that a perpetual novelty exists in the system.

The ant model – MANTA by Drogoul and Ferber [2], generic ecosystem model ECHO by Hraber et al. [12] and Food web simulator – GHEKO by Schmitz et al. [11] are examples of multiagent simulators. Usually a multiagent simulator has a target to reproduce some emergent patterns of behavior. For instance, MANTA was used for the study of labor separation in ant societies. ECHO was used for the study of abundance and distribution of species emerging in the process of Darwinian evolution. We are using our simulator LEM ( Logic for Ecological Modeling) (Krivov, [6]), to study route selection dynamics among Indian peafowl .

Route selection dynamics in an ecosystem is inherent in the multi-factorial nature of interaction of animals among themselves as well as with their environment. Many animal behavior patterns pertaining to their spatial movement occur on a periodic daily basis that are regular and are regulated by internal biological clocks of animals. It would be interesting to find out how the adult animals as well as their offspring react to the environmental factors and how the route selection process emerges. Very few efforts have been made to develop such simulation models. Route selection dynamics is one of the complex problems that zoologists, ethologists and animal ecologists are encountering in their field studies. "Route selection" is a complex phenomenon in nature that arises when the resources are limited or widely distributed within a spatially heterogeneous environment.

2. LEM – Tool for Ecological Modeling Implemented in Visual Prolog

2.1 LEM characteristics

The overall structure of LEM models, syntax of model specification, language, and the functionality of LEM modules are now available on Internet at the URL: http://geocities.com/SiliconValley/Cable/3109/ OR http://www1.50megs.com/lem/

OR http://www.50megs.com/www1/lem/ site includes a provisional manual for LEM users:

LEM is a forward chaining rule-based system implemented in Visual Prolog (Krivov et al. [7]). The behavior of agents and reaction of environment on the actions of agent are specified by the sets of production rules. LEM models could be simple or evolutionary, time may be linear or cyclic. In evolutionary models, each agent has its own rule base that also serves as a set of chromosomes. In simple ecological models, the specie of agents shares the rule base and it should be preprogrammed. There is a special rule table that defines the reactions of the environment to the actions of agents. LEM supports interactive development of individual based ecological models. There is a special language for specification of models. No knowledge of computer languages is required for development of new models, and it seems that LEM is the first general-purpose individual-based modeling tool that could completely relieve ecologists and artificial life researchers from programming work.

LEM runs under Windows 95/98/NT. It could be ported to OS/2. It has professional quality visual interface that is comparable with the best examples of Windows programming. LEM is designed and implemented with the use of Object Oriented Paradigm. It has high performance. On Pentium 133 MHz, 32 Mb RAM LEM could handle population of 1000 –10000 agents with a reasonable speed (5-0.2 steps per one second). LEM comes with the set of tools that facilitate development of individual based models: Text editor with a set of standard Windows functions; fast Data Base system that could store the relevant elements of simulation either in RAM or on disk; Data Base query system; Plotter and Pattern tracer- a special tool for the monitoring of the evolution of system organization and emergence of novel organization. Pattern tracer directly uses the concepts of system organization.

Following are the characteristics of LEM:

Ecosystem: LEM ecosystem is a 2 dimensional grid of sites. It might be similar to ecosystem design in Echo, Gecko, and Manta. However, sites in LEM ecosystem have normal 2 dimensional topology. Each agent of LEM has double representation of its location. Agents know the site, where it is located; in addition, it knows its’ absolute position in the frame of reference of the ecosystem. The sites of ecosystem store information about agents, which are currently located there, plus information about actions of agents. Monitor is the part of LEM that displays the positions of agents on the screen. The_Law rule table of LEM plays the role of Fight Monitor of Gecko. Agents: Modeler could specify the behavior of LEM agents, number of their states and some other parameters in run time. LEM is object oriented forward chaining rule based system and thus has greater flexibility than simulators designed with traditional object oriented methodology. Evolution: The current implementation of LEM support study of behavioral evolution that can happen within in a relatively short time. Simulation: LEM was designed as a general-purpose tool for ecological modeling. Finally, it should be used for simulation of real ecosystems. All other simulators except some like SWARM does not pursue this objective. Database: Almost all multiagent simulators have some tools for monitoring the simulation. However, LEM database has unmatchable capacity for performing this task. Pattern Tracer: Pattern tracer is unique LEM tool for tracing possible patterns of interaction within ecosystem.

2.2 Models of Adaptation supported by LEM

The current version of LEM supports 2 models of adaptation – Genetic Adaptation and Route Adaptation

Genetic adaptation

Evolutionary models of LEM mimic natural evolution. Genetic adaptation is the most common to ALife simulators' way of adaptation. In evolutionary models any etho-agent possesses its private set of schema that are treated as chromosomes during the process of reproduction. Thus in this model of evolution only the schemata of agents are evolving. The evolution of schemata mimics evolution of agent’s behavior.

Signature based evolutionary models allow to study evolution of the physical properties of agents. Although this option is implemented in the current version of LEM, it will not be discussed in the present paper.

Route adaptation

Cyclic models of LEM which were designed during our study of Peafowl’s daily migration mimic spatial and temporal adaptation of animals to the changing environment This is the kind of adaptation that is taking place during the life of individual. The gist of route adaptation is as follows:

Each agent has in memory the preferred route. Route is a collection of triples <DayTime, Place Weight>, Here DayTime is a member of the DayTimeList specified in the term dayandyear (see Krivov [6]) for the description of model specification terms); Place specifies some location in the ecosystem; Weight is an integer that shows how much the agent likes to be in the given place Place at the time DayTime .
The migration of agent according to its route is invoked when agent makes action follow_route. When this action is executed, the agent follows with high speed to the location that corresponds to the current time in agent’s route.
By default the weight of the route-slots for the current time slightly decrements at every step. This means that by default agents have a tendency to forget their route. However there are some actions that could reinforce positively or negatively the weight of the current place in case it is a part of the route
If the weight of some slot becomes less than 0, the slot should be removed from the route.
If during the stepwise decrement of the route-slots the slot for the current time is missing, then new slot should be created for the present time and location. This slot will have some intermediate value of weight.

The route adaptation model should serve for study of spatial behavior. It may be relevant for the study of environmental impact assessments when effect of environmental changes on spatial distribution of animals or birds is to be evaluated.

3. Local Migration of Avi-Fauna in JNU Ecosystem

3.1 General Field Survey

Jawaharlal Nehru University Campus lies on the off-shoots of northern part of the Aravali hills locally known as Delhi Ridge in southern Delhi (28 deg 12’ – 28 deg 53’ north latitudes and 76 deg 50’ – 77 deg 23’ east longitudes; average altitude of Delhi is 213-219 meters above sea level. The campus is spread over an area of 412.54 hectares on South Central Ridge. This land belongs to the category of `tropical thorn forest’. More than 50% of this land is covered with vegetation and has no major permanent water bodies; hence is dependent on monsoons for its water supply. The quartzite rocks form the base of the JNU campus. The soil cover on the lower areas is formed of weathered material. Tropical thorn forest is spread over this soil cover under semi-arid climatic conditions. Large parts of this land is now occupied by man made-structures .The mainland or macrohabitat is made of Dense vegetation area, seen in the lower parts of the Campus which consists of thick base of soil cover holding fast growing and drought resistant perennial species; Woodland like area i.e. the areas covered by widely spaced trees, or other vegetation and Scrubland that covers more than fifty percent of the total Campus area. It includes perennial drought resistant varieties of shrubs and bushes.

A survey was conducted on the Avi-fauna of the campus, and a checklist of birds was prepared (Dahiya and Krivov, [1]). From this list, Indian peafowl was selected as the animal model of the study. Knowledge about ecology and behavior of any peacock pheasant in the wild is very limited. This makes it difficult to establish a satisfactory regular pattern of field routine. The most fundamental field method used was the SURVEY WALKS. The area was walked along fixed paths and locations of peafowl were plotted on the gridded map of the study area, at set times for several consecutive days and the observations were made accordingly.

3.2 Indian Peafowl, Pavo cristatus was selected mainly for the following reasons:

1. This specie could be easily recognized, especially the male, by its large size, iridescent colors, erect crest with unique display and its distinct loud call which could be easily heard and recognized over long distances.

2. Peafowl is gregarious in nature and flocks could be recognized which usually return to their resting sites every night.

3. The bird maintains its territory within limited space though home range may vary according to seasons.

4. Migration is generally limited to daily-migration for foraging if its habitat is not affected. Moreover it is sedentary and terrestrially adapted bird.

    1. Habitat selection made by peafowl

The observations supported with other relevant studies (cited as per the context) show following criteria in habitat selection made by the peafowl that influence its local migration:

1. Availability of water is their main requirement (Johnsingh and Murali [3]). For the present study, location of water bodies, and their relationship with the feeding or resting space is yet under observation.

2. Tall trees to roost on favorably near feeding space, dense bushes with open areas for breeding, and availability of adequate nest sites and feeding grounds (Johnsgard, [13]).

3.As peafowl is omnivorous and usually a primary consumer, the criteria of habitat selection based on availability of food is equally important. Its preference of feeding spaces is in the order of : containing grains> human oriented food> natural food.

Peafowl maintains its home-range and is a lek-breeding species (Rands et al., [9]; Ridley et al. [10]) and shows philopatry to their display site (tends to roost and rest near to display sites which are few meters apart over the whole breeding season) (Rands et al. [9]; Petrie et al. [8]). Such leks are situated apart from each other in woodland and more clumped in open land (Hillgarth [4]). Peafowl is normally attached to its traditional roosting site and usually returns to it on sunset. This site is near its feeding space and also the water source. It selects tall trees for roosting, especially the tree which contains the first branch considerably at a higher distance from the ground which helps escape from predators. In the present study, the primary roosting site was found around the three traditional feeding grounds. This area also contains higher density of peafowl population.

The field observations made in JNU campus on the habitat use by peafowl revealed some set patterns of its daily migration. The space is divided into 18 resting spaces and 21 feeding spaces (see screenshots a and b) in accordance with the movement of associations of peafowl. Each association could be of many flocks and some flock may behave differently in their movement and may explore entirely new feeding or resting space in the real ecosystem. This aspect we are keeping for our more complex models in order to make the present model simple. Likewise, 18 major associations were observed which generally move to feeding spaces at sunrise and return to its resting spaces at sunset.

Peafowl’s daily migration is regulated by photoperiod. In the present field study it was observed that peafowl prefers seeds and other such food provided by the local residents in the feeding spaces nos. 1, 3 & 4 (see screenshots- fig. 1 a&b) which cites they visited regularly almost at the same time at sunrise and at sunset and would wait in the surroundings nearby if the human feeders were late. Some observations made on individual birds/flock suggest that peafowl has a very sharp biological clock. One hen and sometimes the whole flock was observed crossing the road between resting space number 12 to the feeding space number 14 exactly around 5.30 pm in the evening, which observation coincides with the clearing of Hostel mess (Godavari Hostel) and preparations for dinner. One peacock was observed moving from resting space no. 18 to feeding space no. 19 around 7.00 am in the morning and re-entering to the same resting space during 7.30 pm in the evening. One flock was seen moving and foraging during 6.30 a.m. daily near the backside of School of International Studies, for few months repeatedly.

4. Simulation of Peafowl Migration Using LEM

4.1 Specifications

The model is simple (non-evolutionary) and cyclic. It is cyclic for the simple reason that time is cyclic; it is divided into days and years, while day is subdivided into periods. Specification of model is given in a file with extension md. The specification contains pointers to different files that comprise the model. Complete model specification requires one md file, which specifies the model; one lw file, which specify the law of interactions for the current model. It may also include one or several esp files, which specify different species of etho-agents; one prs file specifies all producers that may be included in the model; one ens file that specifies all types of environmental agents in the model.

In peafowl model we use the symbols for different species: Food types: "w"- grasses; "m"-herb; "F" small shrub; "T"- tree; "&"-peafowl.

For peafowl model the Law table that define the outcomes for all interactions between these three species is as following:

species([["&","c"],["w","m","F","T"],["tree","roadv","build"]])

rule([obj(2,"w"),action(1,"&",act2("eat",2))],s([re(1,income(10)),re(2,loss(10))]))

rule([obj(2,"m"),action(1,"&",act2("eat",2))],s([re(1,income(10)),re(2,loss(10))]))

rule([obj(2,"F"),action(1,"&",act2("eat",2))],s([re(1,income(10)),re(2,loss(10))]))

rule([obj(2,"T"),action(1,"&",act2("eat",2))],s([re(1,income(10)),re(2,loss(10))]))

rule([action(1,"c",act2("atack",2)),obj(2,"&"),far(1,2)],s([re(1,bad),re(2,ok)]))

rule([action(1,"c",act2("atack",2)),obj(2,"&"),near(1,2)],s([re(1,income(100)),re(2,death)]))

rule([action(1,"c",act2("atack",2)),obj(2,"&")],case(0.3,[re(1,income(100)),re(2,death)],[re(1,bad),re(2,ok)]))

rule([action(1,"&",any)],s([re(1,ok)]))

rule([action(1,"c",any)],s([re(1,ok)]))

In this table numbers are used as variables for agents, for example the rule:

rule([obj(2,"w"),action(1,"&",act2("eat",2))],s([re(1,income(10)),re(2,loss(10))]))

could be interpreted in natural language as, "if object 1 of specie "&" (peafowl) performs an action "eat" on object 2 of specie "w" then they get the reaction in terms of income or loss. In this example object 1 gets an income of 10 and 2 a loss of 10.

The routes defined for the agents (placed in file having extension ag) are following:

route("rt1",[s("night","lm1"),s("sunrise","lm3"),s("morning","lm5"),s("early_noon","lm5"),s("late_noon","lm7"),s("sunset","lm3")])

route("rt2",[s("night","lm8"),s("sunrise","lm7"),s("morning","lm6"),s("early_noon","lm4"),s("late_noon","lm6"),s("sunset","lm7")])

ag_route("&","rt1")

ag_route("&","rt1")

ag_route("&","rt2")

ag_route("&","rt2")

The species "&" (peafowl) is defined in the following table. Number of states is defined by the predicate number_of_states(3). Peafowl has three state variables, each state variable have double representation as integer and symbol. The state map defines the transformation between the two representations. The first state in which the peafowl could be at one instance of time are either "high", "good", "normal", "low", "bad", "dead". The second state could be either "neutral", "sexy". The third state that represents the age of peafowl has symbolic values: "baby", "adult", "matured" or "dead". The predicate speed(6) specifies the average speed of motion. The duration of pregnancy (gestation + hatching) of peafowl and the number of offspring are specified by the predicate pregnancy(600,3). The predicate schema specifies the behavior of peafowl and staterule specify numeric state updates.

name("&")

number_of_states(3)

state(1,[s(10000,15000,"high"),s(5000,10100,"good"),s(3000,5100,"normal"),s(1000,3100,"low"),s(0,500,"bad"),s(-100,1,"dead")])

state(2,[s(0,50,"neutral"),s(49,70,"sexy")])

state(3,[s(0,600,"baby"),s(550,1200,"adult"),s(1100,7000,"matured"),s(6700,12000,"dead")])

shape("&",0x00FF0000)

speed(6)

pregnancy(600,3)

route_update(1,3000)

schema(1,[obj(1,"c")],act2("runfrom",1))

schema(2,[dtimeto(7),state(3,"baby"),hasleader],follow_leader)

schema(3,[dtimeto(7)],follow_route)

schema(4,[mysx(m),season("summer"),my_state_g(1,"&",3),sm_sx(1,"&")],act2("threaten",1))

schema(4,[mysx(m),season("summer"),my_state_l(1,"&",3),sm_sx(1,"&")],act2("runfrom",1))

schema(4,[season("summer"),state(3,"matured"),state(2,"sexy"),opt_sx(1,"&")],act2("mating",1))

schema(9,[state(1,"high")],act1("moverandom"))

schema(5,[obj(1,"w")],act2("eat",1))

schema(5,[obj(1,"T")],act2("eat",1))

schema(6,[obj(1,"m"),state(1,"normal")],act2("eat",1))

schema(7,[obj(1,"m"),state(1,"bad")],act2("eat",1))

schema(8,[obj(1,"F"),state(1,"bad")],act2("eat",1))

schema(9,[],act1("moverandom"))

staterule(any,act2("eat",1),income(10),[d(0,100),d(1,10),d(2,1),d(3,1)])

staterule(any,act2("mating",1),ok,[d(0,100),d(1,-8),d(2,-40),d(3,10)])

staterule(any,act2("runfrom",1),ok,[d(0,-200),d(3,1)])

staterule(any,any,any,[d(1,-1),d(2,3),d(3,1)])

In specification of schema, numbers are used as variable names and also refer to states of agent. For example schema(4,[season("summer"),state(3,"matured"),state(2,"sexy"),opt_sx(1,"&")],act2("mating",1))

uses number 2 to refer to the second state of the agent state(2,"sexy") ) and to refer some agent which is situated at the same site opt_sx(1,"&"). The schema contains left-hand side and right hand side. If LHS is true then the action specified in RHS is performed. Here, in the RHS act2("mating",1) the variable 1 refer to the agent referred in LHS by variables with the same name "1".

The staterule contains signatures and defines modification of agents state variables depending on action of the agent and reaction of the environment of this action. For example, staterule(any,act2("eat",1),income(10),[d(0,100),d(1,10),d(2,1),d(3,1)])

states that if an agent with any signature agent performs action "eat" with reaction income(10) it gets updates of state 1 by 10,state 2 by 1, state 3 by 1, while d(0,100) specifys route reinforcement for the current route slot.

The map of the Jawaharlal Nehru University (JNU) Campus is plotted on the lattice used for all these models (screen shots in figs.1. a and b). Ecosystem is defined in file named JNUpeafowl.md having dimensions ecosystem(120,96,20) that represent 412.54 hectares (43992894.314 sq. feet) of the land marked by the ecosystem. The day-length of 24 hours was divided into 6 intervals.

lem-sc1.gif (64695 bytes)

lem-sc2.gif (66453 bytes)

(The abbreviations used in two screenshots are: Academic complex: SLS - School of Life Sciences, SC&SS - School of Computer and System Science, SSS - School of Social Studies, SES - School of Environmental Studies, SIS - School of International Studies, SL - School of Language, Literature and Culture studies, Lib – Library, BIC - Bioinformatic center, Admn - Administrative block; Prk - Parthasarthy rock; Prg - Parthasarthy rock ground; Prd1 - Parthasarthy rock depression number 1 (facing road); Prd2 - Parthasarthy rock depression number 2 (facing Prk); Kvs1- School named Kendriya Vidyalaya inside JNU campus; Kc - Kamal Complex shopping center; Gh - Godavari girls hostel; Sh - Sabarmati hostel; Oth - Open Theatre; OTS - Open theatre surroundings; Fs - Feeding space; Rs - Resting space.)

In this model, each association performs some action, which is time and space specific. Four categories of actions have been taken into account, moving, eating, resting, and mating. All these actions are goal-oriented. These actions in the real ecosystem may in turn be bound by many related actions. For instance "moving" means that some association is moving from one space to another for some goal (like feed, forage or roost). LEM supports wide range of actions pertained to moving.

5. Observations and Discussion

Offspring follow their parents for some time and then either get their own routes or move randomly. This route acquiring capacity is decreased in offspring after the increment in generations as per the observations made (figs. 2 a&b).

Figure 2 (a). The population growth curve over a period of two years in approximately 4510 time steps with reproduction steps set at 600 steps and the initial population as w = 89, m = 52, F = 23, T = 26, & = 6, c = 0.

Figure 2(b). The population growth curve over a period of two years in approximately 5490 time steps with reproduction steps (in the specification of "&") set at 600 steps and the initial population as w = 89, m = 52, F = 23, T = 26, & = 6, c = 0.

Figure 3 (a) The screenshot of the prototype model of peafowl’s local migration at the 0-time step

Figure 3 (b) The screenshot after the birth of offspring on completion of one whole year.

The observations in Fig. 2(a) show the population growth curve at 4510 time steps when reproduction is completed in just 6 steps and fig. 2 (b) shows the population growth curve over a period of two years in approximately 5490 time steps with reproduction steps set at 600 steps and the initial population as w = 89, m = 52, F = 23, T = 26, & = 6, c = 0. Seasonal dynamics may be noted here as "&" is preprogrammed to reproduce only during summers. The rise of population is periodic. The offspring are able to inherit sometime whole route from the parents, or with reduced number of periods and sometimes having no period at all. Accordingly, offspring’s route is defined. In Fig. 3 (a) shows a screenshot at the 0-time step and Fig. 3 (b) shows after the birth of offspring in on completion of one whole year.

This model presents a good example of artificial migration and artificial biological-rhythm and its dynamics that arise out of movement of "&" from one point to another which is based on real observations. Many animal behavior patterns occur on a periodic daily basis that is regular. Biological rhythms are manifestations that are regulated by internal biological clocks of animals. These rhythms could be dependent upon the environmental factors or a self–sustaining and adjustable pacemaker mechanism could be set up in accordance with the feed back from the environment. In nature, animals show different rhythms with periods ranging from few minutes to several years. An important observation noted by us is the fact that often offspring develop their own routes. However, with the rise of population the route selection also lowers down. Since the model of real ecosystem is in progress valuable, observations would come out after some more efforts that are in progress.

The route adaptation model should serve for study of spatial behavior. It may be relevant for study of environmental impact assessments when effect of environmental changes on spatial distribution of animals has to be evaluated.

Acknowledgements

The grants provided by Council of Scientific and Industrial Research (CSIR) , New Delhi India to Dr. Dahiya are greatly acknowledged. We are grateful to Carsten Cristoffersen from Prolog Development Center (PDC) Denmark for timely sending a complementary copy of Visual Prolog compiler. Outstanding features of Visual Prolog 5.1 helped Dr. Krivov to implement the design of LEM in a very short time. Special thanks to Logic Programming Association (LPA), England for a complementary copy of LPA Prolog compiler that was used for prototyping and design of LEM. We would like to express our thanks to all the members of Nature Club ARANYAK, registered with WWF, New Delhi. We are grateful to Peter Van Wonterghem, Ph.D. Scholar of Jawaharlal Nehru University who lent us his laptop for typing this report

References

Journals & Proccedings:

[1] Dahiya, Anju. and Serguei. Krivov (1999 March-April issue). Checklist of the birds of JNU campus, Jawaharlal Nehru University New Delhi India. JNU News. Vol. XVII No. 2: pp. 13-16.

[2] Drogoul, Alexis and Jacques Ferber (1992). "Multi-agent simulation as a tool for modeling societies: application to social differentiation in ant colonies". Lecture notes in artificial intelligence, Cristiano Castelfranchi, Eric Werner (Eds.). Artificial Social Systems 4th European workshop on modeling autonomous agents in a multi-agent world, MAAMAW’92, S. Martino al Cimino, Italy. Springer Verlag, 830, pp. 3-23.

[3] Johnsingh A.J.T. and S. Murali. 1978. The ecology and behavior of the Indian Peafowl (Pavo cristatus) of Injar. J. Bombay Natural History Society. 75: 1069-79.

[4] Hillgarth N. 1984. Social organization of Peafowl in India. WPA Journal, 9: 47-56.

[5] Krivov, Serguei, Anju Dahiya and Jaweed Ashraf (1998). "Ethological Principles and Architecture for Adaptive Autonomous Agents". In: proceedings of The 4-Th International Conference On Cognitive Systems, New Delhi, India. Vol. 2.: pp. 604-614.

[6] Krivov, Serguei (1999). A logic based framework for artificial Life. Ph.D. Thesis (submitted). The Inter-cultural Open University, The Netherlands.

(Web site: http://geocities.com/SiliconValley/Cable/3109/serguei.html

[7] Krivov, Serguei, Anju Dahiya and Jaweed Ashraf (1999). "LEM - Multi Agent Simulator Implemented In Visual Prolog". In: proceedings of The 5-Th International Conference On Cognitive Systems, New Delhi, India.

[8] Petrie M. 1992. Peahens prefer Peacocks with elaborate trains. Animal Behavior, 41: 323-32.

[9] Rands M.R.W., M.W. Ridley and A.D. Lelliot. 1984. The social organization of feral peafowl. Animal Behavior, 32: 830-35.

[10] Ridley M.W., A.D. Lelliot and M.R.W. Rands. 1984. The coutship display of feral peafowl. WPA Journal, 9:57-68

[11] Schmitz, O.J. and G. Booth. (1996). "Modeling food web complexity: the consequence of individual based spatially explicit behavioral ecology on trophic interactions". Evolutionary Ecology (in press).

Books:

[12] Hraber, Peter T, Terry Jones and Stephanie Forrest (1996). "The Ecology of ECHO". Artificial Life (in Press).

[13] Johnsgard P.A. 1986. Pheasants of the world. Oxford University Press. London.