Merry E. Makela

AN ARTIFICIAL INTELLIGENCE MODELLING APPROACH TO SIMULATING ANIMAL/HABITAT INTERACTIONS

Ecological modellers have begun to recognize the potential of object-oriented programming techniques in structuring models. However, little has been done to take advantage of artificial intelligences (AI) symbolic representations to model the decision-making processes of animals. Here, a generic model of animal-habitat interaction and a specific model of moose-, Alces alces L., forest interactions in Finland are described that are event-driven and behavior-based. Individual level simulation is accomplished through an object-oriented knowledge representation scheme and AI techniques to implement a hierarchical decision-making model of behavior. The habitat is likewise represented in an object-oriented scheme, allowing the simulation of a heterogeneous environment. Other AI techniques for modelling behavior, memory, and actions are discussed including LISP methods, rule-based reasoning, and several search algorithms. Simulations of the moose-forest system show the power of this approach but are not intended to advance the theory of large-herbivore behavior and foraging. AI techniques are found to be most beneficial in (a) studying population processes based on individual level models of behavior, (b) modelling spatial heterogeneity, (c) building event-driven models, (d) providing a conceptual clarity to model construction, and (e) providing a structure equally well suited to simulating resource management.

Object-oriented simulation: plant growth and discrete organ to organ interactions

Abstract This paper reviews and applies new hierarchical approaches to ecological modelling. These new approaches are made possible by the development of the object-oriented paradigm. This paradigm draws upon the notion of ‘universal’ or classes dating back to early Greek philosophy. It is an intriguing approach to simulation because it is based upon the concepts of hierarchy and taxonomy, two of the basic organizing principles in ecology. Adopting an object-oriented approach to simulation can result in a reduction of mathematical and statistical abstraction. The object-oriented approach lends itself directly to incorporation of mechanisms within appropriate hierarchies. A case study is presented outlining the design steps for simulating plant growth objects (roots, stem, leaves, fruit, whole plant, etc.). The design steps are shown in graphical form to illustrate the differences between object-oriented and traditional procedural approaches. Cotton growth and development has been selected for the case study because of the large knowledge base available for the explicit representation of age and size for each organ. Inclusion of mechanisms at the level of the individual organ provides additional information for crop management. Computing fruit growth at discrete branch locations results in the ability to manage for optimum fiber yield and reduced pest vulnerabilities for individual bolls. Variability in light interception, leaf age, and resulting carbohydrate supply for leaves at specific positions leads naturally to variability in individual boll fiber yield. The goal of the model to capture the behavior of individual organs as a function of their interaction with other organs was achieved. The object-oriented paradigm facilitates the formulation of a simulation procedure in which an individual organ interacted with other organs and the crop microclimate. This led us one step closer to answering the question, ‘How do individual characteristics and behaviors result in given population patterns?’.

Accuracy of numerical methods for solving the advection–diffusion equation as applied to spore and insect dispersal

Abstract Three algorithms for solving a simplified 3-D advection–diffusion equation were compared as to their accuracy and speed in the context of insect and spore dispersal. The algorithms tested were the explicit central difference (ECD) method, the implicit Crank–Nicholson (ICN) method, and the implicit Chapeau function (ICF) method. The three algorithms were used only to simulate the diffusion process. A hold-release wind shifting method was developed to simulate the wind advection process, which shifts the concentration an integer number of grids and accumulates the remaining wind travel distance (which is less than the grid spacing) to the next time step. The test problem was the dispersal of a cloud of particles (originally in only one grid cell) in a 3-D space. The major criterion for testing the accuracy was R 2 , which represents the proportion of the total variation in particle distribution in all grid cells that is accounted for by the particle distribution through numerical solutions. Other criteria included total remaining mass, peak positive density, and largest negative density. High R 2 values were obtained for the ECD method with (Δ t K z )/(Δ z ) 2 ≤0.5 (Δ t =time step; K z =vertical eddy diffusion coefficient; Δ z =vertical grid spacing), and for the two implicit methods with Δ t K z /(Δ z ) 2 ≤5. The ICN method gave higher R 2 values than the ICF method when the concentration gradients were high, but its accuracy decreased more rapidly with the progress of time than the ICF method with a combination of a large grid spacing and a large time step. With very steep concentration gradients, the ICF method generated huge negative values, the ICN method generated negative values to a lesser extent, and the ECD method generated only small negative values. It was also found that good mass and/or peak preservation did not necessarily correspond to a higher R 2 value. Based on the R 2 value and the requirement for concentration positivity, for simulations with steep concentration gradients, the ECD method would be most appropriate, followed by the ICN method, and the ICF method would be least appropriate due to large negative values. For simulations with low concentration gradients, the ECD or ICF or ICN method could be used, but the ICN method would not be appropriate for use in a combination of a large time step and a large grid spacing. The results from this study could help selection and use of appropriate numerical methods in studying the spatial dynamics of spores and insects.

An object-oriented intracolonial and population level model of honey bees based on behaviors of European and Africanized subspecies

Abstract A computer-based simulation model of a population of honey bee colonies, based on the biology of European and African races of Apis mellifera L., is described. The intracolonial dynamics of each colony are simulated each day based on parameters internal and external to the colony. The honey bee population is the collection of all colonies in the model at a particular point in time. Since most of the model development centers on structure and dynamics of the individual colony, the majority of the discussion will be at the intracolonial level. Nevertheless, there is no logical limit to the number of colonies that can be simulated simultaneously. Colonial birth and death processes cause the addition and deletion of colonies to the model as needed. A sample simulation is presented to illustrate both the intracolonial and population level dynamics.

Generation of mechanistic variability in a process-based object-oriented plant model

Abstract A cotton crop model based on individual plant developmental behavior and variability was developed. Object-oriented simulation (OOS) provided the conceptual basis for the new model structure. The procedural model, COTSIM, provided the theoretical background for cotton plant development. Data collected during 1987 from field-grown cotton were used for model development and verification, and data from 1988 were used for model validation. The model predicted mass accretion and production of organs within the patterns and magnitudes observed in the field. The model also predicted crop development aspects that had not previously been described by procedural models. Age and size of leaves and fruit and associated developmental variability were included in the model through representation of objects and their variable behavior defined by their position on the plant and how this constrains their growth. Observed variability was the result of the aggregate behavior of components. Variability in our OOS model is an output as opposed to being an input in most procedural plant models. The model has recreated both realistic plants and populations in a mechanistic simulation. Object-oriented models are an important step towards common structures and languages for model design and the development of simulations. It was noted that increased mechanistic detail resulted in an increase of procedure calls (messages) and a five-fold increase in model run time.

Inclusion of plant structure and fiber quality into a distributed delay cotton model to improve management and optimize profit

Abstract This research introduces a simple variation of distributed delay algorithms to solve some relevant problems in simulation of plant development. Cotton was used as a case study. An energy-based model of cotton plant growth was expanded to account for production of fruit at different main-stem nodes and fruiting branch positions. The inclusion of plant structure in a simulation model permitted more accurate estimation of projected harvest value. The implications to crop and pest management are two-fold. Higher levels of resolution in the plant structure can result in improved estimates of crop value and insect economic injury levels. More explicit representation of plant structure provides a more natural framework for integrated insect models based on organism behavior such as location and fruit size preference. An algorithm is presented to enhance the resolution of ecological applications that include developmental variability. The effects of cotton plant structure on cotton lint yield and fiber quality were determined for a short-season cultivar, TAMCOT CD3H. An existing cotton plant model, COTSIM, was used and adapted to short-season production. The cotton plant model was modified to include plant architecture. Fiber quality parameters and yield of fruit at different branch positions were determined for the test cultivar. Arrays corresponding to different fruiting branch positions and fiber property trends were constructed to better corresponde with patterns observed in field data. These patterns showed that fiber properties are closely associated with the positions on a fruiting branch. The development of fruit position-cohorts was modeled using an algorithm that simulates the distribution of growth rates with a time-distributed delay. This algorithm was extended to include arrays to represent fruit produced at different fruiting branch positions on the plant. The model predicted observed fruiting data from non-stressed plants well. The inclusion of the effects of plant structure on cotton yield and fiber quality permitted a more accurate determination of cotton prices. In addition, when linked to insect pest models, this model will simulate a dynamic, position-dependent effect of insect damage on fruiting structures and economic loss.

Analysis of an algae-based CELSS. I - Model development

A steady state chemical model and computer program have been developed for a life support system and applied to trade-off studies. The model is based on human demand for food and oxygen determined from crew metabolic needs. The model includes modules for water recycle, waste treatment, CO2 removal and treatment, and food production. The computer program calculates rates of use and material balance for food. O2, the recycle of human waste and trash, H2O, N2, and food production supply. A simple non-iterative solution for the model has been developed using the steady state rate equations for the chemical reactions. The model and program have been used in system sizing and subsystem trade-off studies of a partially closed life support system.