Methodology for Holarchic Ecosystem Model based on Ontological tool

In this paper, we present a methodology for aquatic ecosystem modeling. We based our model on the holarchic (non directional hierarchic) nature of the ecosystems. Aquatic ecosystems are crossed by structuring fluxes (light radiation, mass transport, etc.) which confer multi-level of emergent organizations. Elsewhere, a part of aquatic ecosystems are species organizations called food chain. These ones are themselves interacting complex subsystems with multi non-directional retro-actions between them and their environment. These retroactions can be modeled at different levels. Thus, we propose an hybrid holarchic compartmental model. This one aims at easing the gathering of phenomenon in a multi-model, multi-level simulation for studying food chain. We present a methodology based on the use of an ontological tool, called Protegé 2000 and allowing an interface with biological specialists to describe animal behaviors with a hierarchical and structured approach. Elsewhere, the ontology is also based on accurate description of all the species of emergent organizations (homogeneous structures or complex functional systems) that we have to manage dynamically during the simulation. The implementation is build on ProActive package which manages dynamic distribution of simulations over networks, using code migration, for example.Results of first simulations using this environment are shown. We so propose a constructive methodology for concrete simulations over distributed computations, allowing to study real problems in all their complexity, specifically in multi-scale description. 1 ECOSYSTEMS ARE COMPLEX The botanist Tansley (Tansley, 1935) defines ecosystems as ”The more fundamental conception is ... the whole system (in the sense of physics) including not only the organismcomplex but also the whole complex of physical factors forming what we call the environment. We cannot separate them (the organisms) from their special environment with which they form one physical system ... It is the system so formed which [provides] the basic units of nature on the face of the earth ... These ecosystems, as we may call them, are of the most various kind and sizes.” Typically, ecosystems are described as a biotope and a biocenisis in mutual interaction (S.E.Jorgensen and Müller, 2000). Moreover, they are crossed by fluxes (mass transport or energy) which dynamically structure them. A reductionist approach fails in modeling such mutual interactions and feed-back processes. New approaches based on general system theory concepts have therefore been tried to produce more efficient models. Ecosystems are systems as described in the general system theory (Le Moigne, 1994) and can be seen as a set of interacting elements which are characterized by following aspects: • mutual dependence. Each element is directly linked with other elements in structure or dynamic. Therefore, its evolution depends on the other elements in interaction with it. Finally, separating an element from its neighborhood modifies it. • emergence of organizations. The interaction of elements leads to the emergence of natural organizations which generate “new entities”. Those entities differ from their components in their structure and dynamic. • feed-back processes. This is the retroaction from the natural organizations to its own components. Emergence processes act recursively and generate hierarchical systems organization. An adapted description is the one called SOHOS according to Koestler (Koestler and Smythies, 1969). SOHOS stands for Self-Organized Holarchic Open Systems. An holarchy is a non-directional hierarchy in which the members are called holons. If we consider the three previous characteristic aspects of ecosystems, an adapted definition of ecosystems could be: “biotope+biocenosis”, natural multi-level holarchic systems, crossed by structuring fluxes. Ecosystems as holarchic systems, must be studied at many levels of time and space (each level is significant) and are crossed by fluxes. Many tools exist to model and simulate ecosystems. Those tools differ by the nature of the model they use, the level they describe and the phenomena they directly take into account. Thus, they mainly focus on one aspect of the ecosystem. Therefore, gathering all these models is a hard task due to structural differences between these tools. Mixing these models is seldom theoretically feasible, and when it is, the computation of the resulting simulation is often crippled by a terrible computation time. Unfortunately, raising the number of interactions modeled is the concern of most modelers. We aim at providing a model which would facilitate the gathering of models for simulating aquatic ecosystems in a multi-scale description. 2 AN HOLARCHIC HYBRID MODEL FOR ECOSYSTEMS First, let’s start with a description of an ecosystem life cycle. An ecosystem always go through three states characterized by its biotope and its inner complexity. • To begin with, the profile of an ecosystem can be described as juvenile. The ecosystem itself contains many raw materials and its biodiversity is low. The biotope is mainly made of simple organisms. Those organisms modify the layout as they keep on multiplying and consuming the raw material. The environmental condition has little influence on their development. The main factor limiting their growth in size and number is the quantity of material and space. • Then, an ecosystem evolves into a mature or adult form. An adult ecosystem undergoes a replacement of its original settlement (simple organisms) with complex organisms. Those organisms suit the ecosystem characteristics and tend to maintain them. They consume and produce (or participate to the production of) raw materials. Those organisms replace the simple organisms (but some simple organisms remain). So on, the biodiversity of the ecosystem is very high. We should be aware that the complex organisms need the building made by the simple organisms during the juvenile stage to appear. • Finally, an ecosystem tends to be aging. When being adult, the ecosystem is made of many types of complex systems, each one participating to the maintenance of the system. As it grows old, the ecosystem tends to lower his complexity and biodiversity by eliminating the less performing complex organisms and conserving the best association of organisms. During its life, an ecosystem always undergoes stress period making it pass from one step to another (sometimes regressing as it is the case when an ecosystem is being exploited). We point out that biotope and biocenosis are directly linked and that each organism acts for his ecosystem as the same time as it is influenced by it. Following this line of thought, we adopt a well-known description of the biotope based on a classification of the organisms. That classification is made of three parts: 1. Producers. Producers are the base of the ecosystem. They are responsible for the production of organic mass by consuming inorganic materials (mineral salt for example). Moreover, they produce many substances and heat. They too are eaten by consumers. 2. Consumers. An ecosystem most of the time is studied through its organic mass. In an ecosystem, most organisms try to develop and support its organic mass. That category of organism is consumers. They mostly produce organic mass by consuming other organic mass and release part of that mass in the ecosystem. They too participate to the layout of the environment . 3. Detritivors or decomposers. As an organism lives, it releases (dead) organic mass. That organic mass is reused by bacterias to produce raw materials needed by the producers. They thus maintain the ecosystem resources. Each part of that classification is a part of the functional aspect of the ecosystem and participate to the previous cycle of life. That classification constitutes the base of the representation of the biotope in our model. We represent ecosystems with a multi-level model. Some level are fixed “a priori”. We shall describe them later. The determination of the level lays on the following concepts. Ecosystems are systems as described in the general system theory (Bertalanffy, 1968) meaning they are made of elements in interaction. In ecology, the basic elements are individuals. So on, we should introduce individuals in our model. Clearly, we should use an individual based model on that level. Moreover, ecosystems are thermodynamical systems. We should take into account flows between ecosystems on a different level than the individual’s one (as they operate on a different scale). Finally, those flows are well modeled by laws.Finally, ecosystems are SOHOS. That induces we shall define non-directional relations between levels and furthermore, and add dynamic scaling for emerging entities. While studying the informations available concerning several natural aquatic ecosystems (the Seine estuary, for example) we noticed that these ecosystems could be separated into different compartments, each one being a particular ecosystem. Clearly, that description is space-oriented depending of the localization of the compartment and its inner space. A level corresponding to those compartment should be adopted. Definition of the holarchic hybrid model Following the previous concepts, we propose a model suiting ecosystems features and allowing reuse of already existing model different by nature (law or rule-based, different scales). The model is individual-based in the way it represents the different entities of an ecosystem. As a consequence, each entity has states (e-states, i-states or p-states) and behaviors. The behaviors correspond to the model used to model the dynamic of the entity. It could be rule-based or law-based. Our model presents three levels defined “a priori”. • Individual level. That level is the lower one in the holarchy of our system. It embodies entities that could not be decomposed. We introduce individuals and super-individuals at that level. Moreover, the elements of that level are described following the consumer-producer-detritivor model. Thus, one should question what are the links between a particular entity and the ecosystem. That level is clearly individual-based so behaviors of the entities are rulebased and entities have i-states. • Compartment level. The space is a critical data in ecosystem modeling. We introduce the space at that level . Thus, a compartment is a single entity in interaction with other compartment (exchanging flows) and containing individuals. As ecosystems are SOHOS, individuals and compartments influence each other. The compartments have law-based behaviors and e-states. The e-states of the compartment correspond to global values considered homogeneous in all the space of the compartment. • System level. Systems are non-spacialized entity corresponding to a set of entity. The link between those entities could be defined “a priori” or during the simulation. The first example of a system is the ecosystem itself. It is made of compartments and defines their relations. Moreover, phenomena occur at the scale of the ecosystem itself influencing all the compartments. The systems have e-states and law-based behaviors. During the simulation, different kind of systems could emerge, each one corresponding to a new scale. Moreover, phenomena with particular scales should be modeled through systems. It is important to understand that the phenomena occurring on the ecosystems level directly modify the compartment and thus influence too the individuals of the compartment themselves. Simultaneously, individuals modify their compartment and thus influence the ecosystem. What will be modeled We tested our model on a classic example of aquatic ecosystem. We will now describe it. Our example is an ecosystem in which we study the influence of light and oxygen on a simple food chain. The light differs in many place of the ecosystem. The oxygen influences the behavior of the biotope. The food chain is made of four species. The first population is made of planktons consuming mineral salt and producing oxygen depending on the light. Bacterias constitute the second one. They decompose organic mass and release mineral salt. The two last populations are fishes. The first of them feeds on planktons. The second have plankton and the first population of fish as preys. Both of them consume oxygen. 3 AN ONTOLOGY TO DESCRIBE