This review grew out of our realization that models play an increasingly important role in conservation but are rarely used in the research of most avian biologists. Modelers are creating models that are more complex and mechanistic and that can incorporate more of the knowledge acquired by field biologists. Such models require field biologists to provide more specific information, larger sample sizes, and sometimes new kinds of data, such as habitat-specific demography and dispersal information. Field biologists need to support model development by testing key model assumptions and validating models. The best conservation decisions will occur where cooperative interaction enables field biologists, modelers, statisticians, and managers to contribute effectively.

We begin by discussing the general form of ecological models-heuristic or mechanistic, "scientific" or statistical-and then highlight the structure, strengths, weaknesses, and applications of six types of models commonly used in avian conservation: (1) deterministic single-population matrix models, (2) stochastic population viability analysis (PVA) models for single populations, (3) metapopulation models, (4) spatially explicit models, (5) genetic models, and (6) species distribution models. We end by considering the intelligent use of models in decision-making, which requires understanding their unique attributes, determining whether the assumptions that underlie the structure are valid, and testing the ability of the model to predict the future correctly.

Esta revisión surgió al reconocer que los modelos juegan un papel cada vez más importante en conservación, pero son raramente usados en las investigaciones realizadas por la mayoraía de los biólogos que trabajan con aves. En la actualidad se están creando modelos complejos que involucran mecanismos que podrían incorporar más del conocimiento que han adquirido los biólogos de campo. Estos modelos requieren que los biólogos de campo provean información más específica, utilicen tamaños muestrales mayores y que en algunos casos provean nuevos tipos de datos, como demografía en hábitats específicos e información sobre dispersión. Los biólogos de campo deben apoyar el desarrollo de modelos a través de la prueba de los supuestos claves y la validación de los modelos. Las mejores decisiones en conservación ocurrirán al existir una interacción cooperativa y efectiva entre biólogos de campo, biólogos que realizan modelos, estadísticos y personas que trabajan en manejo.

Comenzamos discutiendo la forma general de los modelos ecológicos-heurísticos o que describen mecanismos, "científicos" o estadísticos-y luego destacamos la estructura, las fortalezas y debilidades y las aplicaciones de seis tipos de modelos que se utilizan comúnmente en la conservación de aves: (1) modeles determinísticos matriciales de una única población, (2) modelos de análisis estocásticos de viabilidad poblacional (AVP) para una única población, (3) modelos metapoblacionales, (4) modelos espacialmente explícites, (5) modeles genéticos y (6) modelos de la distribución de las especies. Terminamos considerando el use inteligente de modelos en la toma de decisiones, le que requiere entender los atributos específicos de cada modelo, determinar si los supuestos que subyacen a la estructura son válides y probar la habilidad del modele para predecir el futuro correctamente.

INTRODUCTION

AVIAN BIOLOGISTS INVOLVED in conservation activities encounter formal mathematical and simulation models ever more frequently and in ever more diverse forms. Models are constructed to act as descriptions of ecological systems (Maynard Smith 1974). As in other sciences, such models have driven the development of certain concepts in conservation biology, such as population viability and metapopulation dynamics. Mathematical and simulation models (hereafter "models") have been used to predict outcomes based on past, current, or projected conditions, and they serve a useful role in synthesizing knowledge and guiding research. To make a model, one is forced to state explicitly the relations between external factors and the state of the system, and this quickly reveals the limits of our understanding. More significantly, models have become important tools that are applied to policy decisions, and their use will continue to expand as desktop computing power grows and user-friendly software makes modeling increasingly accessible. Models, however, are neither a panacea nor the only useful kind of analysis for making conservation decisions. Intelligent use of models in decision-making requires understanding their unique attributes, determining whether the assumptions that underlie the structure are valid, and testing the ability of the model to predict the future correctly.

The present review, and a symposium at an American Ornithologists' Union (AOU) meeting sponsored by the AOU Conservation Committee, grew out of our realization that models play an important role in conservation but are rarely incorporated in the research of most avian biologists. For example, at a recent AOU meeting, only -4% of 317 papers presented or tested models, compared with -21% at a meeting of the Ecological Society of America held a few days earlier. Nonetheless, most presenters at both meetings employed a statistical model to test the significance of, or evaluate patterns in, their data. Talking about models with ornithologists evokes strong reactions, as evidenced by the responses of AOU meeting attendees to the question: "What is the first thing that you think of when I say the words 'model or ecological model'?". Answers included " hot air," "money for someone else," "predicting the future," "I go right to the Discussion and hope that they know what they are doing," "people who haven't been in the field enough," "computers," "assumptions and generalizations," "reality?," and "something I don't understand at all."