Interpreting Bird Atlas Data Using Logistic Models: An Example From Lesotho, Southern Africa

Osborne, Patrick E.; Tigar, Barbara J.

doi:10.2307/2404347

Cited by 108 publications

(71 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…These models are usually fitted using generalised linear modelling (McCullogh and Nelder, 1989) or generalised additive modelling (Hastie and Tibshirani, 1990). For example, logistic regression has been used widely to predict the occurrence and habitat use of endangered vertebrate species (Ferrier, 1991;Lindenmayer et al, 1991;Mills et al, 1993;Pearce et al, 1994;Mladenoff et al, 1999), game species (Straw et al, 1986;Diffenbach and Owen, 1989) and vascular plants (Austin et al, 1990), to examine the response of species to environmental perturbation (Reckhow et al, 1987), and to model the regional distributions of a large number of fauna and flora species to provide information for regional forest conservation planning (Osborne and Tigar, 1992;NSW NPWS, 1994a,b).…”

Section: Introductionmentioning

confidence: 99%

Evaluating the predictive performance of habitat models developed using logistic regression

Pearce

Ferrier

2000

Ecological Modelling

1,676

1,425

View full text Add to dashboard Cite

The use of statistical models to predict the likely occurrence or distribution of species is becoming an increasingly important tool in conservation planning and wildlife management. Evaluating the predictive performance of models using independent data is a vital step in model development. Such evaluation assists in determining the suitability of a model for specific applications, facilitates comparative assessment of competing models and modelling techniques, and identifies aspects of a model most in need of improvement. The predictive performance of habitat models developed using logistic regression needs to be evaluated in terms of two components: reliability or calibration (the agreement between predicted probabilities of occurrence and observed proportions of sites occupied), and discrimination capacity (the ability of a model to correctly distinguish between occupied and unoccupied sites). Lack of reliability can be attributed to two systematic sources, calibration bias and spread. Techniques are described for evaluating both of these sources of error. The discrimination capacity of logistic regression models is often measured by cross-classifying observations and predictions in a two-by-two table, and calculating indices of classification performance. However, this approach relies on the essentially arbitrary choice of a threshold probability to determine whether or not a site is predicted to be occupied. An alternative approach is described which measures discrimination capacity in terms of the area under a relative operating characteristic (ROC) curve relating relative proportions of correctly and incorrectly classified predictions over a wide and continuous range of threshold levels. Wider application of the techniques promoted in this paper could greatly improve understanding of the usefulness, and potential limitations, of habitat models developed for use in conservation planning and wildlife management.

show abstract

Section: Introductionmentioning

confidence: 99%

Evaluating the predictive performance of habitat models developed using logistic regression

Pearce

Ferrier

2000

Ecological Modelling

1,676

1,425

View full text Add to dashboard Cite

show abstract

“…However, even a perfect knowledge of the biology of a species cannot guarantee that a static map will reflect dynamic properties of species distribution (Tyre et al, 2001). Moreover, human and logistic limitations make it impractical to survey large areas and, inevitably, our knowledge of the spatial distribution of most species will have many gaps (Osborne and Tigar, 1992). A common solution to this problem is to resort to predictive habitat modelling (reviewed in Guisan and Zimmermann, 2000, and see some recent examples in Osborne et al, 2001;Scott et al, 2002;Seoane et al, 2003) and regard the results as potential habitat, able to be reached and colonised by a species (Tyre et al, 2001).…”

Section: Introductionmentioning

confidence: 99%

Competing roles for landscape, vegetation, topography and climate in predictive models of bird distribution

Seoane

Bustamante

Dı́az-Delgado

2004

Ecological Modelling

View full text Add to dashboard Cite

Predictive habitat models rely on the relationship between a response variable (either occurrence or abundance of a species) and a set of environmental predictors. Vegetation is generally preferred as a source of potential predictors because of having a more direct link with reproductive necessities of species than topography and climate. However, vegetation maps are costly to produce and update and most land-use/land-cover maps are usually made with a general purpose, focused on land management, and not thinking on animal distribution. On the contrary, basic topographic and climatic maps are easier to obtain or to derive, are not so sensitive to legend design, and do not need of such frequent updates. In this study, we compare the predictive ability of different sets of environmental predictors when modelling the distribution of breeding birds. Models were generated for 79 bird species in South-western Spain using Generalised Additive Models (GAMs) with binomial errors and logit link. For each species, several models were created that differed in the set of candidate predictors initially tested (either derived from vegetation or from topo-climatic maps) or in the conditional order in which those predictor sets were tested. Within vegetation predictors, a similar strategy was used to ascertain the relative relevance of vegetation landscape (variables describing the surrounding habitat matrix) compared to vegetation cover (variables describing the type of vegetation found on sampling sites).Vegetation models were significantly more accurate than topo-climatic models, but the difference was due to the higher number of potential predictors in the set of vegetation variables. Vegetation landscape models were significantly more accurate than vegetation cover models, even when controlling for the number of candidate predictors. Models that included both sets of predictors (vegetation and topo-climatic variables) had a slightly superior predictive ability. Our results indicate that, when building predictive models of bird distribution, the best results are obtained using both vegetation and topo-climatic variables as potential predictors. If time or budget constrains compel to concentrate on a single set of predictors, selection should be done on grounds of data availability, because model accuracy is likely to be similar for models derived from vegetation compared to topographic and climatic predictors. In relation to vegetation predictors, vegetation landscape reflects important information not revealed by vegetation cover measures at the sampling site; thus, regional modelling programmes would certainly gain predictive ability by including landscape patterns that are currently disregarded.

show abstract

“…For this purpose, we use the South African avifauna as a test case. We do this because there are marked spatial variations in the abiotic environment across South Africa, specifically a marked east-west aridity gradient (Schulze 1997a;see also O'Brien 1993see also O'Brien , 1998O'Brien et al 1998), and because it has long been maintained that vegetation complexity accounts for most variation in bird species richness in the region (Winterbottom 1978;Osborne and Tigar 1992;Allan et al 1997).…”

mentioning

confidence: 99%

Species Richness, Environmental Correlates, and Spatial Scale: A Test Using South African Birds

Rensburg¹,

Chown²,

Gaston³

2002

The American Naturalist

View full text Add to dashboard Cite

show abstract

Interpreting Bird Atlas Data Using Logistic Models: An Example From Lesotho, Southern Africa

Cited by 108 publications

References 7 publications

Evaluating the predictive performance of habitat models developed using logistic regression

Evaluating the predictive performance of habitat models developed using logistic regression

Competing roles for landscape, vegetation, topography and climate in predictive models of bird distribution

Species Richness, Environmental Correlates, and Spatial Scale: A Test Using South African Birds

Contact Info

Product

Resources

About