TESTING ECOLOGICAL THEORY USING THE INFORMATION‐THEORETIC APPROACH: EXAMPLES AND CAUTIONARY RESULTS

By S. Richards. 2005

573 citations recorded by the source corpus.

Subfields: Ecological Modeling, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/3f658ebf9a13f464523cea82ae964813da34a313

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

In Nature and Landscape Conservation

Citations in the corpus

Modeling Tree Diameter Distributions in Natural Forests: An Evaluation of 10 Statistical Models (2015, unknown): “The use of larger thresholds avoids models providing good descriptions to data being discarded due to small sample sizes (when n is small, models with fewer parameters tend to perform better [Richards 2005]).”
Habitat selection models for European wildcat conservation (2008, methodology): “Fitted models were compared and hierarchically ordered using Akaikes Information Criterion (AIC), a statistic method that rewards parsimony by penalizing the maximum likelihood for the number of model parameters (Akaike, 1973; Richards, 2005).”
The rarity of direct experiences of nature in an urban population (2017, unknown): “Following Richards (2005) we retained all models where Δ AIC c 6.”
Combining vessel-based surveys and tracking data to identify key marine areas for seabirds (2009, background): “Nevertheless, Richards (2005) raised some concerns related to model averaging with Akaike weights and stressed the need for continued research on the effectiveness of model averaging, as did Burnham & Anderson (2004).”
Quantifying the effects of prey abundance on killer whale reproduction (2009, methodology): “Posterior probabilities and AIC model weights were calculated for each model as a measure of data support (Burnham & Anderson 2002), but these quantities were not used for model averaging because of potential biases they might introduce in predictions ( Richards 2005 ).”
A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion (2010, methodology): “2010), and criticism of AIC is likewise prevalent (e.g. Guthery et al. 2005; Richards 2005; Stephens et al. 2005; Link and Barker 2006).”
Changing philosophies and tools for statistical inferences in behavioral ecology (2009, methodology): “Behavioral Ecology doi:10.1093/beheco/arp137 Advance Access publication 2 November 2009 Review Changing philosophies and tools for statistical inferences in behavioral ecology”
Using information theory as a substitute for stepwise regression in ecology and behavior (2010, unknown): “…comparable and relatively high support, then choosing a permissive cutoff criterion may lead to uncertain results (i.e., very divergent confidence sets) that strongly depend on the cutoff value, while a strict cutoff will lead to the use of most models in the candidate set (also see Richards 2005).”
Multimodel inference in ecology and evolution: challenges and solutions (2011, methodology): “Although many biologists agree with the principles behind using this approach, the ways and means of applying a multimodel procedure and model averaging to various types of biological problems are still in their infancy (see also Richards, 2005).”
When are eradication campaigns successful? A test of common assumptions (2012, background): “This threshold of 6 is much higher than the widely-accepted rule-of-thumb of selecting all models with a DAIC value B2 proposed by Burnham and Anderson (2002), but it recently has been demonstrated that the most parsimonious model, defined as the one with the lowest expected Kullback– Leibler distan”
A general biodiversity–function relationship is mediated by trophic level (2017, methodology): “We defined the best model set as all models with δaic < 2 (Richards 2005), Burnham and Anderson 2002).”
Multiple facets of diversity effects on plant productivity: Species richness, functional diversity, species identity and intraspecific competition (2019, background): “Although a 259 commonly adopted rule of thumb states that a model with ΔAIC<2 is likely to be the best model 260 (Burnham & Anderson, 2002; Richards, 2005), here we doubled this threshold to reduce the 261 probability of having false positive.”
Model averaging in ecology: a review of Bayesian, information-theoretic, and tactical approaches for predictive inference (2018, methodology): “In simulations, AIC weights did not reliably identify the model with the known lowest KL-divergence or prediction error (Richards 2005, Richards et al. 2011).”