How to assess the prediction accuracy of species presence–absence models without absence data?

By Wenkai Li, Qinghua Guo. 2013

98 citations recorded by the source corpus.

Subfields: Ecological Modeling, Ecology, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/56baa28547c9243bed7175b280a15c04dd6694be

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Ecological Modeling

1. Matt D. White
2. G. Newell
3. Canran Liu
4. C. Meynard
5. Morgane Barbet-Massin
6. Boris Leroy
7. Ming Xu
8. M. Gillard

In Ecology

1. Matt D. White
2. G. Newell
3. Canran Liu
4. C. Meynard
5. Morgane Barbet-Massin
6. Boris Leroy
7. Ming Xu
8. M. Gillard

In Nature and Landscape Conservation

1. Matt D. White
2. G. Newell
3. Canran Liu
4. C. Meynard
5. Morgane Barbet-Massin
6. Boris Leroy
7. Ming Xu
8. M. Gillard

Citations in the corpus

• The area under the precision‐recall curve as a performance metric for rare binary events (2019, methodology): “When observed absences are not available, several authors have advised using performance metrics that do not incorporate specificity (Hirzel, Le Lay, Helfer, Randin, & Guisan, 2006; Li & Guo, 2013).”
• Climate change‐induced decline in bamboo habitats and species diversity: implications for giant panda conservation (2015, methodology): “As AUC is not appropriate to evaluate the accuracy of binary predictions, we also used true skill statistic (TSS) as suggested by recent studies (Lobo et al., 2008; Li & Guo, 2013) to assess the accuracy of the bamboo species models.”
• On the selection of thresholds for predicting species occurrence with presence‐only data (2015, methodology): “Li and Guo (2013) derived a statistic Fpb (which will be defined in the next section) based upon the presencebackground (which is defined as random points) data acting as a surrogate for the F measure (i.e., the harmonic mean of positive predicted value and sensitivity), and used it to select…”
• Climate Change Threatens Giant Panda Protection in the 21st Century (2015, methodology): “Areas under the Operating Characteristic Curve (AUC) is a widely-used approach to evaluate model performance of species distribution models, but it is a threshold-independent measure that should not be applied to binary predictions (Lobo et al., 2008; Li and Guo, 2013).”
• Present and future distribution of three aquatic plants taxa across the world: decrease in native and increase in invasive ranges (2017, methodology): “To achieve this objective, species distribution models (SDMs) can be used as tools for the early detection of invasive species (Broennimann et al. 2007; Guisan et al. 2013), especially when establishing predictive scenarios, since such models have generally been proven to reflect the correct…”
• Without quality presence–absence data, discrimination metrics such as TSS can be misleading measures of model performance (2018, methodology): “Our simulations unequivocally indicate that when sample prevalence is different from species prevalence, none of the tested metrics are independent of species prevalence, corroborating previous conclusions on the TSS (Somodi et al., 2017), and invalidating the propositions of Li and Guo (2013).”