#### DMCA

## Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation

### Cached

### Download Links

Citations: | 114 - 2 self |

### Citations

2379 | Generalized Additive Models
- HASTIE, TIBSHIRANI
- 1990
(Show Context)
Citation Context ... interesting topic for future research would be to develop new feature classes that allow even more succinct response approximations, such as the splines commonly used in generalized additive models (=-=Hastie and Tibshirani 1990-=-). The logistic output format, introduced here, is easier to interpret than previous Maxent output formats: it can be interpreted as estimating a species probability of presence, conditioned on enviro... |

1309 | The elements of statistical learning: data mining, inference and prediction - Hastie, Tibshirani, et al. |

1125 |
Information theory and statistical mechanics
- Jaynes
- 1957
(Show Context)
Citation Context ...ince the set of constraints typically under-specifies the model, among all probability distributions satisfying the constraints, we choose the one of maximum entropy, i.e. the most unconstrained one (=-=Jaynes 1957-=-). Maximum entropy density estimation can also be explained from a decision theoretic perspective as robust Bayes estimation. Specifically, consider the scenario where the goal of the modeler is to op... |

942 | Greedy function approximation: a gradient boosting machine - Friedman |

505 | On discriminative vs. generative classifiers: A comparison of logistic regression and naive Bayes - Ng, Jordan - 2002 |

490 |
Elements of information theory, 2nd ed
- Cover, Thomas
- 1990
(Show Context)
Citation Context ...equence of independent samples from the Maxent distribution ql, corresponding to a sequence of observations. Then the average of their log probabilities will be very close to H, the negative entropy (=-=Cover and Thomas 2006-=-), because H is simply the mean log probability: / H Sql (x) ln (ql(x)): Thus, for ‘‘typical’’ sites whose log probabilities are close to this mean, we obtain ql(x):e H . The model Q therefore assigns... |

456 | Novel methods improve prediction of species’ distributions from occurrence data - Elith, Graham - 2006 |

432 |
A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ. Conserv
- Fielding, Bell
- 1997
(Show Context)
Citation Context ...mulative and logistic formats are all monotonically related, so they rank sites in the same order and therefore result in identical performance, when measured using rank-based statistics such as AUC (=-=Fielding and Bell 1997-=-). However, their predictive performance will vary when measured by statistics that depend on actual output values such as Pearson’s correlation (Zheng and Agresti 2000). Experimental methods Species ... |

379 | Maximum entropy modeling of species geographic distributions. Ecological Modelling 190 - Phillips, Anderson, et al. - 2006 |

298 |
Multivariate adaptive regression splines (with discussion
- Friedman
- 1991
(Show Context)
Citation Context ...ps et al. 2006), which are used to model piecewise constant responses. An analogous step up from piecewise constant responses to piecewise linear responses has been helpful in the regression setting (=-=Friedman 1991-=-). The second extension is a new logistic output format, which addresses the fact that the existing raw and cumulative formats can be hard to interpret. For many modeling applications, we are interest... |

288 | Extinction risk from climate change. Nature - Thomas, Cameron, et al. - 2004 |

140 | Conservatism of ecological niches in evolutionary time - Peterson, Soberón, et al. - 1999 |

113 | Game theory, maximum entropy, minimum discrepancy, and robust bayesian decision theory
- Grnwald, Dawid
- 2002
(Show Context)
Citation Context ...gy which guarantees the best performance regardless of p, also called the minimax strategy, is to choose the maximum entropy distribution subject to the given constraints (Topsøe 1979, Grünwald 2000, =-=Grünwald and Dawid 2004-=-). To understand how p represents the realized distribution of the species, consider the following (idealized) sampling strategy. An observer picks a random site x from the set X of sites in the study... |

102 |
Bayesian regularization and pruning using a Laplace prior
- Williams
- 1995
(Show Context)
Citation Context ...at comparison. There are a number of other factors that may help explain Maxent’s good performance. First, Maxent uses l1 regularization, which tends to produce models with few non-zero coefficients (=-=Williams 1995-=-, Tibshirani 1996) and therefore encourages parsimony. Regularization appears to prevent overfitting better than variable-selection methods commonly used for regression-based models such as generalize... |

97 | A maximum entropy approach to species distribution modeling - Phillips, Dudík, et al. - 2004 |

80 | Performance guarantees for regularized maximum entropy density estimation - Dudik, Phillips, et al. - 2004 |

62 |
Use and interpretation of logistic regression in habitat-selection studies
- Keating, Cherry
- 2004
(Show Context)
Citation Context ...have not been used for species distribution modeling to date. Perhaps more importantly, when used on presence-only data, regression-based methods suffer from the problem of ‘‘contaminated controls’’ (=-=Keating and Cherry 2004-=-), in 172Table 7. Average performance for current versions and settings of Maxent and boosted regression trees (BRT), evaluated on independent presence/absence test data using area under the receiver... |

58 | Niche-based modelling as a tool for predicting the risk of alien plant invasions at a global scale. Glob Change Biol 11 - Thuiller, DM, et al. - 2005 |

51 | Predicting species spatial distributions using presence-only data: A case study of native New Zealand ferns - Zaniewski, Lehmann, et al. - 2002 |

44 | 2006: Model-based uncertainty in species’ range prediction - Pearson, Thuiller, et al. |

40 | 2006: Are niche-based species distribution models transferable in space - Randin, Dirnböck, et al. |

39 |
Information theoretical optimization techniques
- Topsoe
- 1979
(Show Context)
Citation Context ...t of constraints. The strategy which guarantees the best performance regardless of p, also called the minimax strategy, is to choose the maximum entropy distribution subject to the given constraints (=-=Topsøe 1979-=-, Grünwald 2000, Grünwald and Dawid 2004). To understand how p represents the realized distribution of the species, consider the following (idealized) sampling strategy. An observer picks a random sit... |

38 | Correcting sample selection bias in maximum entropy density estimation - Dudík, Schapire, et al. - 2006 |

38 | Bias, Variance, and Prediction Error for Classification Rules
- Tibshirani
- 1996
(Show Context)
Citation Context ...ty to the presence sites and less to the rest of the sites, i.e. models that best distinguish the presence sites from the background. The second term, regularization (also known as the lasso penalty; =-=Tibshirani 1996-=-) gets larger as the weights lj get larger. Larger weights lj typically mean that the model is more complex and is thus more likely to overfit. Maximizing the difference between log likelihood and reg... |

35 | Quantitative methods for modeling species habitat: comparative performance and an application to Australian plants - Elith - 2002 |

32 | SE: Habitat history improves prediction of biodiversity in rainforest fauna - CH, Moritz, et al. |

28 | Uses and requirements of ecological niche models and related distributional models - Peterson - 2006 |

27 | Predicting species distributions from museum and herbarium records using multiresponse models fitted with multivariate adaptive regression splines. Diversity and Distributions - Elith, Leathwick |

23 | Predicting the potential invasive distributions of four alien plants species in North America - Peterson, Papes, et al. - 2003 |

19 |
Geographical sampling bias and its implications for conservation priorities in Africa
- Reddy, Dávalos
(Show Context)
Citation Context ...d extension addresses the problem of sample selection bias. Occurrence data are frequently biased, for example towards areas easier to access such as areas near roads, towns, airports, and waterways (=-=Reddy and Dávalos 2003-=-). When the bias is large, presence-only models approximate the biased sampling distribution as much as they approximate the species distribution. This can be avoided by having the background sample r... |

17 | Real vs. artefactual absences in species distributions: tests for Oryzomys albigularis (Rodentia: Muridae) in Venezuela - Anderson - 2003 |

15 | Presence-only data and the em algorithm - Ward, Hastie, et al. |

14 | Maximum entropy and the glasses you are looking through - Grünwald |

7 | et al. 2005. Validation of species-climate impact models under climate change - Araújo |

7 | Presence-only data and the EM algorithm. Biometrics - Ward, Hastie, et al. - 2008 |

5 | et al. 2004. New developments in museum-based informatics and applications in biodiversity analysis - Graham |

5 | Maxent software for species distribution modeling (Version 3.3.3k) [Software]. Retrieved from http://www.cs.princeton.edu/~schapire/maxent - Dudík, Phillips, et al. - 2011 |

2 | A maximum entropy approach to species distribution modeling - J, Dudik, et al. - 2004 |

1 | et al. 2001. Evaluation of museum collection data for use in biodiversity assessment - Ponder |