Species-specific traits associated to prediction errors in bird habitat suitability modelling

Seoane, Javier; Carrascal, Luis M.; Alonso, César L.; Palomino, David

doi:10.1016/j.ecolmodel.2004.12.012

Cited by 86 publications

(88 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The model always had a lower predictive power for Palearctic ducks, highly gregarious species, than for Ardeids, the least gregarious species in this study. This is in accordance with a negative impact of gregariousness on the predictive power of the model, an observation already made by Seoane et al (2005). One approach to improve this was to mask out areas were a bird group had never been observed.…”

Section: Discussionsupporting

confidence: 79%

Ecological Modeling of the Spatial Distribution of Wild Waterbirds to Identify the Main Areas Where Avian Influenza Viruses are Circulating in the Inner Niger Delta, Mali

Cappelle¹,

Girard²,

Fofana

et al. 2010

EcoHealth

View full text Add to dashboard Cite

Predicting areas of disease emergence when no epidemiological data is available is essential for the implementation of efficient surveillance programs. The Inner Niger Delta (IND) in Mali is a major African wetland where >1 million Palearctic and African waterbirds congregate. Waterbirds are the main reservoir of Avian Influenza Viruses (AIV). Our objective was to model their spatial distribution in order to predict where these viruses would be more likely to circulate. We developed a generalized linear model (GLM) and a boosted regression trees (BRT) model based on total aerial bird counts taken in winter over 6 years. We used remotely sensed environmental variables with a high temporal resolution (10 days) to predict the spatial distribution of four waterbird groups. The predicted waterbird abundances were weighted with an epidemiological indicator based on the prevalence of low pathogenic AIV reported in the literature. The BRT model had the best predictive power and allowed prediction of the high variability of waterbird distribution. Years with low flood levels showed areas with a higher risk of circulation and had better spatial distribution predictions. Each year, the model identified a few areas with a higher risk of AIV circulation. This model can be applied every 10 days to evaluate the risk of AIV emergence in wild waterbirds. By taking into account the IND's ecological variability, it allows better targeting of areas considered for surveillance. This could enhance the control of emerging diseases at a local and regional scale, especially when resources available for surveillance programs are scarce.

show abstract

Section: Discussionsupporting

confidence: 79%

Ecological Modeling of the Spatial Distribution of Wild Waterbirds to Identify the Main Areas Where Avian Influenza Viruses are Circulating in the Inner Niger Delta, Mali

Cappelle¹,

Girard²,

Fofana

et al. 2010

EcoHealth

View full text Add to dashboard Cite

show abstract

“…Other studies have shown that the breadth of a species' niche relative to the environmental conditions found in the study area as a whole influences model accuracy more than the breadth of a species' niche per se (Seoane et al 2005;Hernandez et al 2006). More marginal species (those that have niches furthest from the average conditions of the study area) have also been shown to be modelled more accurately than less marginal species, probably for similar reasons (Luoto et al 2005;Seoane et al 2005;Carrascal et al 2006;Hernandez et al 2006). We might expect therefore that the accuracy of species distribution models will decrease with increasing niche breadth or habitat tolerance.…”

Section: Introductionmentioning

confidence: 98%

Effect of characteristics of butterfly species on the accuracy of distribution models in an arid environment

et al. 2009

View full text Add to dashboard Cite

Species distribution models show great promise as tools for conservation ecology. However, their accuracy has been shown to vary widely among taxa. There is some evidence that this variation is partly owing to ecological differences among species, which make them more or less easy to model. Here we test the effect of five characteristics of Egyptian butterfly species on the accuracy of distribution models, the first such comparison for butterflies in an arid environment. Unlike most previous studies, we perform independent contrasts to control for species relatedness. We show that range size, both globally and locally, has a negative effect on model accuracy. The results shed light on causes of variation in distribution model accuracy among species, and hence have relevance to practitioners using species distribution models in conservation decision making.

show abstract

“…Second, it has been argued that species that are difficult to detect are likely to have more incomplete occurrence data. This may result in lower accurate distribution models for these species (Seoane et al 2005). In this study, we test whether increasing the length of the point counts makes it possible, above all, to improve the models' explanatory and predictive performance for species with a low detection probability.…”

Section: Introductionmentioning

confidence: 99%

Point count duration: five minutes are usually sufficient to model the distribution of bird species and to study the structure of communities for a French landscape

Bonthoux

Balent

2011

J Ornithol

View full text Add to dashboard Cite

The point count method is very widely used for estimating bird abundances. In studies of bird-environment relationships, the duration used for point counts varies considerably from one study to another. Short counting times may increase the number of false absences while long counting times increase the probability that birds initially absent immigrate during the counting period. This study aims to quantify the effect of point count duration on the performance of bird distribution models and on the estimated structure of the communities. We used a sample of 256 point counts collected in south western France and we compared four count durations (5, 10, 15 and 20 min). After comparing the predictive performance of seven statistical methods, we constructed one GAM (generalized additive model) per counting time to link the presenceabsence data of each species with seven landscape variables. We evaluated the models by examining explained deviance and by comparing predicted and observed values for an independent data using AUC (area under the curve) and TSS (true skill statistics). At the community level, we constructed one CQO (constrained quadratic ordination) per counting time, then we compared the species' scores along the latent environmental variables. For the 21 species studied, the overall performance of GAMs only improves very moderately with a lengthening of the counting time (mean D 2 = 25% for 5 min and 28% for 20 min; mean TSS = 0.40 for 5 min and 0.43 for 20 min), with an increase for some species and a decrease for others. The latent environmental variables of the four CQOs were associated with the same explanatory variables and the scores of the species along the latent variables were highly correlated (between 5 and 20 min, P \ 2.10 -16 , rho = 0.99). In the perspective of building reliable bird distribution models, our results thus show that a point count duration of 5 min is sufficient in temperate regions such as France.

show abstract

Species-specific traits associated to prediction errors in bird habitat suitability modelling

Cited by 86 publications

References 34 publications

Ecological Modeling of the Spatial Distribution of Wild Waterbirds to Identify the Main Areas Where Avian Influenza Viruses are Circulating in the Inner Niger Delta, Mali

Ecological Modeling of the Spatial Distribution of Wild Waterbirds to Identify the Main Areas Where Avian Influenza Viruses are Circulating in the Inner Niger Delta, Mali

Effect of characteristics of butterfly species on the accuracy of distribution models in an arid environment

Point count duration: five minutes are usually sufficient to model the distribution of bird species and to study the structure of communities for a French landscape

Contact Info

Product

Resources

About