## Introduction

Quantifying the relationships between the distributions of species and their abiotic or biotic environments has a long history in ecological research. While understanding where species occur is a fundamental ecological requirement, prediction of occurrence is essential for much conservation and population management. This is particularly the case for endangered species, where knowing what determines distribution is a necessary precursor for schemes to mitigate decline or to create new populations through reintroduction. Given our focus on reporting research with an explicitly applied significance, it is perhaps no surprise that a sizeable proportion of manuscripts received by the *Journal of Applied Ecology* has been concerned with modelling species–habitat relationships and distributions. Early in the 1990s, *Journal of Applied Ecology* published one of the key, seminal papers that introduced significant new methodology as well as novel applications to real conservation problems (Augustine, Mugglestone & Buckland 1993). Since 2000, we have published at least 14 papers that consider species distribution modelling (Table 1), and many others have addressed methodological issues associated with data handling or data collection for modelling . While there has been an ongoing interest in the development and application of modelling approaches, the last 3 years have seen fundamental changes in the methodology involved with this form of modelling, with the development of information-theoretic approaches. These developments have been apparent in published papers, and the six papers in this special profile focus on some of the key issues. Not only do they illustrate how widely new data technologies, such as remotely sensed imagery, have pervaded applied ecology but also they reveal how the philosophy of modelling is changing. Five of the papers utilize information-theoretic approaches that are different to the formal hypothesis-testing approach presented in many past papers. We consider here the significance of these recent changes, how *Journal of Applied Ecology* has responded and how they may shape applied ecology in the future.

Authors | Taxon | Species data | Habitat predictors | GIS | Model | Model assessment |
---|---|---|---|---|---|---|

Collingham et al. (2000) | Weeds | Biological recording | Remote-sensed imagery | Yes | Stepwise logistic regression | Kappa statistics |

Cowley et al. (2000) | Lepidoptera | Transect survey | Mapped habitat data | No | Logistic regression | Kappa statistics |

Milsom et al. (2000) | Birds | Field survey | Mapped habitat data and field survey | No | Generalized linear mixed model (logistic and autologistic) | – |

Manel, Buckton & Ormerod (2000) | Birds Invertebrates | Field survey | Field survey | No | Logistic regression using AIC | – |

Bradbury et al. (2000) | Birds | Field survey | Field survey | No | Log-linear and logistic regression | Threshold classification |

Gates & Donald (2000) | Birds | Biological recording | Mapped habitat data | No | Logistic regression | – |

Jaberg & Guisan (2001) | Bats | Augmented biological records | Mapped habitat data | Yes | Poisson regression | Kappa statistics |

Manel, Williams & Ormerod (2001) | Invertebrates | Field survey | Field survey | No | Logistic regression using AIC | Kappa statistics, ROC plots, jack-knifing |

Pearce et al. (2001) | Mammals, reptiles, birds | Field survey | Remote-sensed imagery | Yes | Generalized additive model (logistic) | Modified Z-test |

Osborne, Alonso & Bryant (2001) | Birds | Field survey | Remote-sensed imagery | Yes | Logistic and autologistic regression | ROC plots |

Suárez-Seoane, Osborne & Alonso (2002) | Birds | Field survey | Remote-sensed imagery | Yes | Generalized additive model | Threshold classification |

Ambrosini et al. (2002) | Birds | Field survey | Field survey | No | Logistic and linear regression with quasi-like lihood | Kappa statistics |

Schadt et al. (2002) | Mammals | Radio-tracking | Remote-sensed imagery | Yes | Logistic regression | ROC plots |

Holloway, Griffiths & Richardson (2003) | Lepidoptera | Field survey | Mapped habitat data | Yes | Rule base | – |

Vaughan et al. (2003) | Mammals | Questionnaire | Land classes and farm census | No | Ordinal logistic regression | Concordance between predicted and observed |

Cabeza et al. (2004) | Lepidoptera | Transect survey | Mapped habitat data | No | Logistic regression | Kappa statistics |

Engler, Guisan & Rechsteiner (2004) | Plants | Biological recording | Mapped climatic and terrain data | Yes | Logistic regression | Kappa and ROC plots |

Jeganathan et al. (2004) | Birds | Bird sign (field survey) | Remote-sensed imagery | Yes | Logistic and autologistic regression | – |

Gibson et al. (2004) | Birds | Bird song (field survey) | Remote-sensed imagery | Yes | Logistic regression and information-theoretic | ROC plots |

Johnson, Seip & Boyce (2004) | Mammals | Radio-tracking | Mapped habitat data | Yes | Logistic regression and information-theoretic | K-fold cross-validation |

Frair et al. (2004) | – | Remote sensing (GPS) | Field survey | Yes | Logistic regression and information-theoretic | ROC plots |

All modelling studies have three basic components: a data set describing the incidence or abundance of the species of interest and a data set of putative explanatory variables; a mathematical model that relates the species data to the explanatory variables; and an assessment of the utility of the model developed in terms of a validation exercise or an assessment of model robustness. Recent publications in *Journal of Applied Ecology* have all followed this basic formula, but each has placed different emphasis on the three components. While these differences reflect the interests of individual authors, they also reflect author responses to the underlying ecology of the species that have been studied and an implicit recognition that assumptions have to be made at all stages in the modelling process.