Strategic directions for agent-based modeling: avoiding the YAAWN syndrome

In this short communication, we examine how agent-based modeling has become common in land change science and is increasingly used to develop case studies for particular times and places. There is a danger that the research community is missing a prime opportunity to learn broader lessons from the use of agent-based modeling (ABM), or at the very least not sharing these lessons more widely. How do we find an appropriate balance between empirically rich, realistic models and simpler theoretically grounded models? What are appropriate and effective approaches to model evaluation in light of uncertainties not only in model parameters but also in model structure? How can we best explore hybrid model structures that enable us to better understand the dynamics of the systems under study, recognizing that no single approach is best suited to this task? Under what circumstances – in terms of model complexity, model evaluation, and model structure – can ABMs be used most effectively to lead to new insight for stakeholders? We explore these questions in the hope of helping the growing community of land change scientists using models in their research to move from ‘yet another model’ to doing better science with models.     

Predicting occurrence of juvenile shark habitat to improve conservation planning

Fishing and habitat degradation have increased the extinction risk of sharks, and conservation strategies recognize that survival of juveniles is critical for the effective management of shark populations. Despite the rapid expansion of marine protected areas (MPAs) globally, the paucity of shark‐monitoring data on large scales (100s–1000s km) means that the effectiveness of MPAs in halting shark declines remains unclear. Using data collected by baited remote underwater video systems (BRUVS) in northwestern Australia, we developed generalized linear models to elucidate the ecological drivers of habitat suitability for juvenile sharks. We assessed occurrence patterns at the order and species levels. We included all juvenile sharks sampled and the 3 most abundant species sampled separately (grey reef [Carcharhinus amblyrhynchos], sandbar [Carcharhinus plumbeus], and whitetip reef sharks [Triaenodon obesus]). We predicted the occurrence of juvenile sharks across 490,515 km² of coastal waters and quantified the representation of highly suitable habitats within MPAs. Our species‐level models had higher accuracy (? ≥ 0.69) and deviance explained (≥48%) than our order‐level model (? = 0.36 and deviance explained of 10%). Maps of predicted occurrence revealed different species‐specific patterns of highly suitable habitat. These differences likely reflect different physiological or resource requirements between individual species and validate concerns over the utility of conservation targets based on aggregate species groups as opposed to a species‐focused approach. Highly suitable habitats were poorly represented in MPAs with the most restrictions on extractive activities. This spatial mismatch possibly indicates a lack of explicit conservation targets and information on species distribution during the planning process. Non‐extractive BRUVS provided a useful platform for building the suitability models across large scales to assist conservation planning across multiple maritime jurisdictions, and our approach provides a simple for method for testing the effectiveness of MPAs.     

The role of agent-based models in wildlife ecology and management

Conservation planning of critical habitats for wildlife species at risk is a priority topic that requires the knowledge of how animals select and use their habitat, and how they respond to future developmental changes in their environment. This paper explores the role of a habitat-modeling methodological approach, agent-based modeling, which we advocate as a promising approach for ecological research. Agent-based models (ABMs) are capable of simultaneously distinguishing animal densities from habitat quality, can explicitly represent the environment and its dynamism, can accommodate spatial patterns of inter- and intra-species mechanisms, and can explore feedbacks and adaptations inherent in these systems. ABMs comprise autonomous, individual entities; each with dynamic, adaptive behaviors and heterogeneous characteristics that interact with each other and with their environment. These interactions result in emergent outcomes that can be used to quantitatively examine critical habitats from the individual- to population-level. ABMs can also explore how wildlife will respond to potential changes in environmental conditions, since they can readily incorporate adaptive animal-movement ecology in a changing landscape. This paper describes the necessary elements of an ABM developed specifically for understanding wildlife habitat selection, reviews the current empirical literature on ABMs in wildlife ecology and management, and evaluates the current and future roles these ABMs can play, specifically with regards to scenario planning of designated critical habitats.     

Simulating spatiotemporal dynamics of urbanization with multi-agent systems—A case study of the Phoenix metropolitan region, USA

Urbanization is a human-dominated process and has greatly impacted biodiversity, ecosystem processes, and regional climate. To understand the socioeconomic drivers of urbanization and project future urban landscape changes, multi-agent systems provide a powerful tool. We develop an agent-based model of urban growth for the Phoenix metropolitan region of the United States, which simulates the behavior of regional authorities, real estate developers, residents, and environmentalists. The BDI (Beliefs-Desires-Intentions) structure is employed to simulate the agents behavior and decision models. The heterogeneity of agents is reflected by adjusting parameters according to the agents’ beliefs, desires and preferences. Three scenarios, baseline, economic development priority and environmental protection, are developed and analyzed. The combination of multi-agent system and spatial regression model is employed to predict the future urban development of the Phoenix metropolitan region. Landscape metrics are used to compare the spatial patterns of the urban landscape resulting from different scenarios in different times. In general, with the rapid urban expansion, the shape of urban patches will become more regular as many of them become coalesced. The spatial analysis of urban development through modeling individual and group decisions and human-environment interactions with a multi-agent systems approach can enhance our understanding of the socioeconomic driving forces and mechanisms of urban development.     

Estimating an homogeneous series of a population abundance indicator despite changes in data collection procedure: A hierarchical Bayesian modelling approach

Abundance indicators are required both to assess and to manage wild populations. As new techniques are developed and teams in charge of gathering the data change, data collection procedures (DCPs) can evolve in space and time. How to estimate an homogeneous series of abundance indicator despite changes in DCP? To tackle this question a hierarchical Bayesian modelling (HBM) approach is proposed. It integrates multiple DCPs in order to derive a single abundance indicator that can be compared over space and time irrespective of the DCP used. Compared to single DCP models, it takes further advantage for abundance estimation of the joint treatment of a larger set of spatio-temporal units. After presenting the general formulation of our HBM approach, it is applied to the juvenile Atlantic salmon (Salmo salar L.) population of the River Nivelle (France). Posterior model checking, using χ² discrepancy measure, do not reveal any inadequacy between the model and the data. Despite a change in the DCP used (successive removals to catch-per-unit of effort), a unique abundance indicator for the 425 spatio-temporal units (site × year) sampled over twenty-four years (1985-2008) is estimated. The HBM approach allows the assessment of precision of the abundance estimates and shows variation between DCPs: a reduction in precision is observed during the most recent years (2005-2008) when only the catch-per-unit of effort DCP was used. The merits and generality of our HBM approach are discussed. We contend it extends previous single DCP models or inter-calibration of two DCPs, and it could be applied to a wide range of specific situations (taxon and DCPs).     

Reconstructing historical marine ecosystems using food web models: Northern British Columbia from Pre-European contact to present

Mass-balance trophic models (Ecopath with Ecosim) are developed for the marine ecosystem of northern British Columbia (BC) for the historical periods 1750, 1900, 1950 and 2000 AD. Time series data are compiled for catch, fishing mortality and biomass using fisheries statistics and literature values. Using the assembled dataset, dynamics of the 1950-based simulations are fitted to agree with observations over 50 years to 2000 through the manipulation of trophic flow parameters and the addition of climate factors: a primary production anomaly and herring recruitment anomaly. The predicted climate anomalies reflect documented environmental series, most strongly sea surface temperature and the Pacific Decadal Oscillation index. The best-fit predator–prey interaction parameters indicate mixed trophic control of the ecosystem. Trophic flow parameters from the fitted 1950 model are transferred to the other historical periods assuming stationarity in density-dependent foraging tactics. The 1900 model exhibited an improved fit to data using this approach, which suggests that the pattern of trophic control may have remained constant over much of the last century. The 1950 model is driven forward 50 years using climate and historical fishing drivers. The resulting ecosystem is compared to the 2000 model, and the dynamics of these models are compared in a predictive forecast to 2050. The models suggest similar restoration trajectories after a hypothetical release from fishing.     

The impact of invasive grasses on the population growth of Anemone patens, a long-lived native forb

Negative impacts of invasive plants on natives have been well documented, but much less is known about whether invasive plants can cause population level declines. We used demographic models to investigate the effects of two invasive grasses on the demography and population growth of Anemone patens, a long-lived native perennial of North American grasslands. Demographic data of A. patens growing in patches characterized by Bromus inermis, Poa pratensis, or native grasses were used to parameterize integral projection models. Models based on both average conditions and those allowing for environmental stochasticity indicate that A. patens is slowly increasing in patches of native grass (lambda = 1.02) and declining in patches of invasive grasses, particularly those dominated by B. inermis (lambda = 0.93). Extinction probabilities indicate that A. patens should persist in native grass patches, but has a much higher probability of extinction in Bromus patches compared to Poa patches. While sensitivity analyses showed that survival had the biggest effect on population growth rates in all habitats, results of a Life Table Response Experiment (LTRE) revealed that slower individual growth rates in patches of invasive grasses contributed the most to the observed reduction in population growth. These results suggest that invasive grasses may cause slow declines in A. patens, despite short-term coexistence, and that controlling B. inermis only would not be sufficient to ensure A. patens persistence.     

Mechanistic understanding of human–wildlife conflict through a novel application of dynamic occupancy models

Crop and livestock depredation by wildlife is a primary driver of human–wildlife conflict, a problem that threatens the coexistence of people and wildlife globally. Understanding mechanisms that underlie depredation patterns holds the key to mitigating conflicts across time and space. However, most studies do not consider imperfect detection and reporting of conflicts, which may lead to incorrect inference regarding its spatiotemporal drivers. We applied dynamic occupancy models to elephant crop depredation data from India between 2005 and 2011 to estimate crop depredation occurrence and model its underlying dynamics as a function of spatiotemporal covariates while accounting for imperfect detection of conflicts. The probability of detecting conflicts was consistently <1.0 and was negatively influenced by distance to roads and elevation gradient, averaging 0.08–0.56 across primary periods (distinct agricultural seasons within each year). The probability of crop depredation occurrence ranged from 0.29 (SE 0.09) to 0.96 (SE 0.04). The probability that sites raided by elephants in primary period t would not be raided in primary period t + 1 varied with elevation gradient in different seasons and was influenced negatively by mean rainfall and village density and positively by distance to forests. Negative effects of rainfall variation and distance to forests best explained variation in the probability that sites not raided by elephants in primary period t would be raided in primary period t + 1. With our novel application of occupancy models, we teased apart the spatiotemporal drivers of conflicts from factors that influence how they are observed, thereby allowing more reliable inference on mechanisms underlying observed conflict patterns. We found that factors associated with increased crop accessibility and availability (e.g., distance to forests and rainfall patterns) were key drivers of elephant crop depredation dynamics. Such an understanding is essential for rigorous prediction of future conflicts, a critical requirement for effective conflict management in the context of increasing human–wildlife interactions.     

Incorporating spatial autocorrelation into cellular automata model: An application to the dynamics of Chinese tamarisk (Tamarix chinensis Lour.)

Spatial autocorrelation (SAC) is frequently encountered in most spatial data in ecology. Cellular automata (CA) models have been widely used to simulate complex spatial phenomena. However, little has been done to examine the impact of incorporating SAC into CA models. Using image-derived maps of Chinese tamarisk (Tamarix chinensis Lour.), CA models based on ordinary logistic regression (OLCA model) and autologistic regression (ALCA model) were developed to simulate landscape dynamics of T. chinensis. In this study, significant positive SAC was detected in residuals of ordinary logistic models, whereas non-significant SAC was found in autologistic models. All autologistic models obtained lower Akaike's information criterion corrected for small sample size (AICc) values than the best ordinary logistic models. Although the performance of ALCA models only satisfied the minimum requirement, ALCA models showed considerable improvement upon OLCA models. Our results suggested that the incorporation of the autocovariate term not only accounted for SAC in model residuals but also provided more accurate estimates of regression coefficients. The study also found that the neglect of SAC might affect the statistical inference on underlying mechanisms driving landscape changes and obtain false ecological conclusions and management recommendations. The ALCA model is statistically sound when coping with spatially structured data, and the adoption of the ALCA model in future landscape transition simulations may provide more precise probability maps on landscape transition, better model performance and more reasonable mechanisms that are responsible for landscape changes.     

Spatial and temporal patterns of habitat use by Eurasian oystercatchers (<Emphasis Type="Italic">Haematopus ostralegus</Emphasis>) in the eastern Wadden Sea revealed using GPS data loggers

Detailed knowledge on species–habitat relationships is of crucial importance for the understanding of processes in marine ecosystems. Being top-predators, birds are important bio-indicators for marine systems. The aim of this study was to elucidate precise information on foraging habitat use and foraging times of oystercatchers (Haematopus ostralegus) on wide tidal flats using global positioning system (GPS) data loggers. The study was conducted to collect hints for the negative population trends in oystercatchers in the Wadden Sea. It is the first time that GPS technique has been used in a shorebird species. Although oystercatchers are known to exhibit foraging site fidelity, a number of individuals visited multiple sites. Foraging trips at night were longer, and the targeted sites were further away than those used during the day. These patterns were likely to be caused by higher risks of clutch predation by avian predators during the day that led adults to reduce their absence to defend their clutches. Our methodological approach enabled the subtle spatio-temporal patterns of habitat use to be determined on a very fine spatio-temporal scale. We suggest further potential studies using GPS data loggers that may help to reveal the reasons for the current declines in oystercatcher populations in the German Wadden Sea.     

Eliciting cognitive processes underlying patterns of human-wildlife interactions for agent-based modelling

Integrating humans in our perception of ecosystems is of critical importance to adequately protect natural resources. This poses the challenge of understanding human decision making in the context of decisions potentially threatening nature's integrity. We developed a spatially explicit agent-based model that simulates commercial whale-watching vessel movements based on a representation of the captains’ decision making process when observing marine mammals in and around the Saguenay-St. Lawrence Marine Park in Québec, Canada. We focus here on the human part of the global model, the submodel of whale movements having been developed and validated independently (Lamontagne, 2009). The objective of this study is to select and validate a model of whale-watching captains’ decision making using the pattern-oriented modelling approach (POM): three models of cognitive heuristics (satisficing, tallying and Take The Best) along with a null model (random choice) were tested. These concurrent decision making models were built upon knowledge extracted from data collected during field investigations, including interviews with whale-watching captains and park wardens, onboard and shore-based observations, and analyses of a multi-year dataset of sampled whale-watching excursions. Model selection is performed by statistically comparing simulated and real patterns of boat trajectories (excursion length), spatial hotspots (kernel home range 50%), and excursion content (species observed, time allocated to different activities). The selection process revealed that the Take The Best heuristic was the best performing model. We used the distribution of the number of whale-watching boats in the vicinity (2000 m) of each vessel as a secondary pattern to validate the ability of each decision making model to reproduce real observations. Given the prevalence of the species attribute in the choice of which whale to observe, the Take The Best heuristic's ability to deal with non-compensatory information partly explains its overall best performance. Moreover, implementation of communication abilities between modelled captains led to the emergence of persistent observation sites in the park, which is a well-known collective spatiotemporal characteristic of the whale-watching industry; thus validating the fundamental assumption that cooperation is an important mechanism behind the pattern of whale-watching boat dynamics. The relatively good performance of the satisficing and tallying heuristics supports both field evidence and literature on bounded rationality in that humans likely use collections of heuristics (adaptive toolbox) to solve decision problems in different contexts. The POM strategy appears suitable to build up an informative ABM regarding the management of human activities in a natural environment so that further developments will be assessed following the same approach.     

Effects of uncertainty and variability on population declines and IUCN Red List classifications

The International Union for Conservation of Nature (IUCN) Red List Categories and Criteria is a quantitative framework for classifying species according to extinction risk. Population models may be used to estimate extinction risk or population declines. Uncertainty and variability arise in threat classifications through measurement and process error in empirical data and uncertainty in the models used to estimate extinction risk and population declines. Furthermore, species traits are known to affect extinction risk. We investigated the effects of measurement and process error, model type, population growth rate, and age at first reproduction on the reliability of risk classifications based on projected population declines on IUCN Red List classifications. We used an age‐structured population model to simulate true population trajectories with different growth rates, reproductive ages and levels of variation, and subjected them to measurement error. We evaluated the ability of scalar and matrix models parameterized with these simulated time series to accurately capture the IUCN Red List classification generated with true population declines. Under all levels of measurement error tested and low process error, classifications were reasonably accurate; scalar and matrix models yielded roughly the same rate of misclassifications, but the distribution of errors differed; matrix models led to greater overestimation of extinction risk than underestimations; process error tended to contribute to misclassifications to a greater extent than measurement error; and more misclassifications occurred for fast, rather than slow, life histories. These results indicate that classifications of highly threatened taxa (i.e., taxa with low growth rates) under criterion A are more likely to be reliable than for less threatened taxa when assessed with population models. Greater scrutiny needs to be placed on data used to parameterize population models for species with high growth rates, particularly when available evidence indicates a potential transition to higher risk categories.     

Parallel agent-based modeling of spatial opinion diffusion accelerated using graphics processing units

In this article, we describe a parallel agent-based model of spatial opinion diffusion that is driven by graphics processing units (GPUs). Modeling opinion exchange and diffusion across landscapes often involves the simulation of large numbers of geographically located individual decision-makers and a massive number of individual-level interactions. This simulation requires substantial computational power. GPU-enabled computing resources provide a massively parallel processing platform based on a fine-grained shared memory paradigm. This massively parallel processing platform holds considerable promise for meeting the computing requirement of agent-based models of spatial problems. In this article, we focus on the parallelization of an agent-based spatial opinion model using GPU technologies. We discussed key algorithms designed for parallel agent-based opinion modeling: including domain decomposition and mutual exclusion. Experiments conducted to examine computing performance show that GPUs provide a computationally efficient alternative to traditional parallel computing architectures and substantially accelerate agent-based models of large-scale opinion exchange among individual decision makers.     

Statistical Estimation and Model Selection of Species-Accumulation Functions

Abstract:  Soberón and Llorente (1993) proposed pure-birth stochastic processes as theoretical models for species-accumulation curves, and these processes have frequently been used to describe the progress of biological inventories. We describe, in algorithmic form, an alternative statistical analysis based on a likelihood approach ( Díaz-Francés & Gorostiza 2002 ) that provides mathematical rigor to the ideas in Soberón and Llorente (1993) and improves the estimation of the models by incorporating the facts that the variance of the error is not constant and that the observations are correlated. Additionally, we used the likelihood ratios between candidate models as an objective procedure for model selection, allowing comparison between the goodness of fit of various models. The software for these statistical methods can now be downloaded off the Internet. We used two examples of butterfly data sets to illustrate the use of the methods and the software.     

A multi-agent simulation to assess the risk of malaria re-emergence in southern France

A multi-agent simulation (MAS) was developed to assess the risk of malaria re-emergence in the Camargue in southern France, a non-endemic area where mosquitoes of the genus Anopheles (Culicidae) live. The contact rate between people and potential malaria vectors, or the human biting rate, is one of the key factor to predict the risk of re-emergence of malaria, would the parasite be introduced in the region. Our model (called MALCAM) represents the different agents that could influence malaria transmission in the Camargue – people, mosquitoes, animal hosts and the landscape – in a spatially explicit environment. The model simulates spatial and temporal variations in human biting rate at the landscape scale. These variations depend on the distribution of people and potential vectors, their behaviour and their interactions. A land use/cover map was used as a cellular-spatial support for the movements of and interactions between mobile agents. The model was tested for its sensitivity to variations in parameter values, and for the agreement between field observations and model predictions. The MALCAM model provides a tool to better understand the interactions between the multiple agents of the disease transmission system, and the land use and land cover factors that control the spatial heterogeneity in these interactions. It allows testing hypotheses and scenarios related to disease dynamics by varying the value of exogenous biological, geographical, or human factors. This application of agent-based modelling to a human vector-borne disease can be adapted to different diseases and regions.     

Forecasting extinction risk with nonstationary matrix models.

Matrix population growth models are standard tools for forecasting population change and for managing rare species, but they are less useful for predicting extinction risk in the face of changing environmental conditions. Deterministic models provide point estimates of lambda, the finite rate of increase, as well as measures of matrix sensitivity and elasticity. Stationary matrix models can be used to estimate extinction risk in a variable environment, but they assume that the matrix elements are randomly sampled from a stationary (i.e., non-changing) distribution. Here we outline a method for using nonstationary matrix models to construct realistic forecasts of population fluctuation in changing environments. Our method requires three pieces of data: (1) field estimates of transition matrix elements, (2) experimental data on the demographic responses of populations to altered environmental conditions, and (3) forecasting data on environmental drivers. These three pieces of data are combined to generate a series of sequential transition matrices that emulate a pattern of long-term change in environmental drivers. Realistic estimates of population persistence and extinction risk can be derived from stochastic permutations of such a model. We illustrate the steps of this analysis with data from two populations of Sarracenia purpurea growing in northern New England. Sarracenia purpurea is a perennial carnivorous plant that is potentially at risk of local extinction because of increased nitrogen deposition. Long-term monitoring records or models of environmental change can be used to generate time series of driver variables under different scenarios of changing environments. Both manipulative and natural experiments can be used to construct a linking function that describes how matrix parameters change as a function of the environmental driver. This synthetic modeling approach provides quantitative estimates of extinction probability that have an explicit mechanistic basis.