A practical overview of transferability in species distribution modeling

Werkowska, Wioletta; Márquez, Ana Luz; Real, Raimundo; Acevedo, Pelayo

doi:10.1139/er-2016-0045

Cited by 65 publications

(89 citation statements)

References 107 publications

(85 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…BIOREG=northern (8) r=-0.05 ns r=-0.03 ns BIOREG=southern (2411) r=0.05** r=0.05** BIOREG=eastern (617) r=0.16*** r=0.17*** BIOREG=western (470) r=0.55*** r=0.61*** NUT=hunting ground (3183) r=0.46** r=0.44** NUT=municipality (273) r=0.17*** r=0.16*** NUT=NUTS3 (50) r=0.52*** r=0.51*** Thus, both the full and the simplified final models are quite equivalents in terms of goodness of fit (Table 4) and predictive performance (Table 5; see also Figure 6) and therefore, and having in mind that the model should be extrapolated and downscaled, we opted to explore more in detail the simplified final model since the predicted performance of simpler models is higher when they have to be transferred (Werkowska et al 2016 The present document has been produced and adopted by the bodies identified above as authors. This task has been carried out exclusively by the authors in the context of a contract between the European Food Safety Authority and the authors, awarded following a tender procedure.…”

Section: Validation Dataset (N) Pearson' R Pearson' R Full Final Modementioning

confidence: 99%

ENETwild modelling of wild boar distribution and abundance: update of occurrence and hunting data‐based models

Acevedo

Croft

Smith

et al. 2019

EFSA Supporting Publications

Self Cite

View full text Add to dashboard Cite

In October 2018 the ENETWILD consortium created suitability maps based on available data on wild boar occurrence at 10 km square resolution and initial version of abundance models based on hunting statistics at NUTS3 and NUTS2 resolution, that were statistically downscaled for MSs to 10x10 km grid squares. This report presents updated suitability map for wild boar presence based on additional occurrence data and new algorithms, and new models based on highresolution hunting yield data for MSs and neighbouring countries. New environmental variables closely associated with wild boar abundance and distribution were also included. Our results showed no consensus for a single best occurrence model: out of those tested, both Maxent and random forest could be considered the best options depending on the choice of assessment metric. Predictions from these models notably disagreed in eastern Europe where data on wild boar occurrence are limited. Despite agreement among models, predictions in the south appeared over-predicted, most likely due to a lack of contrasting absence data. Whilst there remain some methodological adjustments which could be tested, substantial improvement in the prediction from occurrence models relies on further collection of wild boar occurrence data in the east and complimentary data on survey effort in the south. The predictive performance of the hunting yield model was high. Although the incorporation of new data at higher spatial resolution markedly improved predictions, such data is still needed in some regions, ideally coupled with hunting effort, which would allow such estimates to be transformed into reliable densities. Comparison between predictions from the occurrence and hunting yield models showed they were statistically 1 ENETWILD Consortium: www.enetwild.com www.efsa.europa.eu/efsajournal 1 EFSA Supporting publication 2019:EN-1674Modelling wild boar distribution associated, but the strength of that relationship was dependent on the type of occurrence model and the bioregion. These findings are compatible with previous interpretations of the occurrence model, and highlight the relevance of obtaining more accurate data, especially from northern and eastern bioregions in Europe.

show abstract

Section: Validation Dataset (N) Pearson' R Pearson' R Full Final Modementioning

confidence: 99%

ENETwild modelling of wild boar distribution and abundance: update of occurrence and hunting data‐based models

Acevedo

Croft

Smith

et al. 2019

EFSA Supporting Publications

Self Cite

View full text Add to dashboard Cite

show abstract

“…Compared to other fields, such as economics and climate science, projection ensembles were introduced to species distribution modelling relatively recently (Araújo & New, 2007;Thuiller, 2004), but gained popularity since specialized modelling platforms became available-such as the R-package 'biomod2' (Thuiller, Lafourcade, Engler, & Araújo, 2009). Yet, the importance of also varying model complexity in projection ensembles has recently been emphasized by several authors (Boria, Olson, Goodman, & Anderson, 2014;Merow et al, 2014;Werkowska, Márquez, Real, & Acevedo, 2017). A literature study of 125 recent papers employing SDM projections revealed that the most frequently varied step was the emission scenario (63% of cases), followed by the climate models used to estimate future climatic conditions (48% of cases) (Figure 1a, for further information see Appendix S1).…”

Section: Introductionmentioning

confidence: 99%

“…Adding more predictors to a model increases the amount of signal and noise available to SDM algorithms and typically leads to larger numbers of parameters estimated, and thus more complex models (Merow et al, 2014;Werkowska et al, 2017). Adding more predictors to a model increases the amount of signal and noise available to SDM algorithms and typically leads to larger numbers of parameters estimated, and thus more complex models (Merow et al, 2014;Werkowska et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

Model complexity affects species distribution projections under climate change

Brun

Thuiller

Chauvier

et al. 2019

Journal of Biogeography

123

125

View full text Add to dashboard Cite

Aim: Statistical species distribution models (SDMs) are the most common tool to predict the impact of climate change on biodiversity. They can be tuned to fit relationships at various levels of complexity (defined here as parameterization complexity, number of predictors, and multicollinearity) that may co-determine whether projections to novel climatic conditions are useful or misleading. Here, we assessed how model complexity affects the performance of model extrapolations and influences projections of species ranges under future climate change. Location: Europe.Taxon: 34 European tree species. Methods:We sampled three replicates of predictor sets for all combinations of 10 levels (n = 3-12) of environmental variables (climate, terrain, soil) and 10 levels of multicollinearity. We used these sets for each species to fit four SDM algorithms at three levels of parameterization complexity. The >100,000 resulting SDM fits were then evaluated under environmental block cross-validation and projected to environmental conditions for 2061-2080 considering four climate models and two emission scenarios. Finally, we investigated the relationships of model design with model performance and projected distributional changes. Results: Model complexity affected both model performance and projections of species distributional change. Fits of intermediate parameterization complexity performed best, and more complex parameterizations were associated with higher projected loss of current ranges. Model performance peaked at 10-11 variables but increasing number of variables had no consistent effect on distributional change projections. Multicollinearity had a low impact on model performance but distinctly increased projected loss of current ranges. Main conclusions: SDM-based climate change impact assessments should be based on ensembles of projections, varying SDM algorithms as well as parameterization complexity, besides emission scenarios and climate models. The number of predictor variables should be kept reasonably small and the classical threshold of maximum absolute Pearson correlation of 0.7 restricts collinearity-driven effects in projections of species ranges. | 131 BRUN et al.

show abstract

“…of Pages 13 simultaneously achieve high accuracy and precision, even when predicting into novel contexts, will provide maximum utility for decision making [9]. To date, however, tests of transferability across taxa and geographic locations have failed to demonstrate consistent patterns (Figure 1), and a general approach to developing transferable models remains elusive (but see [6,10]). Here, we outline challenges that, if addressed, will improve the harmonization, uptake, and application of model transfers in ecology.…”

mentioning

confidence: 99%

Outstanding Challenges in the Transferability of Ecological Models

Yates

Bouchet

Caley

et al. 2018

Trends in Ecology & Evolution

435

456

View full text Add to dashboard Cite

Predictive models are central to many scientific disciplines and vital for informing management in a rapidly changing world. However, limited understanding of the accuracy and precision of models transferred to novel conditions (their 'transferability') undermines confidence in their predictions. Here, 50 experts identified priority knowledge gaps which, if filled, will most improve model transfers. These are summarized into six technical and six fundamental challenges, which underlie the combined need to intensify research on the determinants of ecological predictability, including species traits and data quality, and develop best practices for transferring models. Of high importance is the identification of a widely applicable set of transferability metrics, with appropriate tools to quantify the sources and impacts of prediction uncertainty under novel conditions. Predicting the UnknownPredictions facilitate the formulation of quantitative, testable hypotheses that can be refined and validated empirically [1]. Predictive models have thus become ubiquitous in numerous scientific disciplines, including ecology [2], where they provide means for mapping species distributions, explaining population trends, or quantifying the risks of biological invasions and disease outbreaks (e.g., [3,4]). The practical value of predictive models in supporting policy and decision making has therefore grown rapidly (Box 1) [5]. With that has come an increasing desire to predict (see Glossary) the state of ecological features (e.g., species, habitats) and our likely impacts upon them [5], prompting a shift from explanatory models to anticipatory predictions [2]. However, in many situations, severe data deficiencies preclude the development of specific models, and the collection of new data can be prohibitively costly or simply impossible [6]. It is in this context that interest in transferable models (i.e., those that can be legitimately projected beyond the spatial and temporal bounds of their underlying data [7]) has grown.Transferred models must balance the tradeoff between estimation and prediction bias and variance (homogenization versus nontransferability, sensu [8]). Ultimately, models that can Highlights Models transferred to novel conditions could provide predictions in data-poor scenarios, contributing to more informed management decisions.The determinants of ecological predictability are, however, still insufficiently understood.Predictions from transferred ecological models are affected by species' traits, sampling biases, biotic interactions, nonstationarity, and the degree of environmental dissimilarity between reference and target systems.We synthesize six technical and six fundamental challenges that, if resolved, will catalyze practical and conceptual advances in model transfers.We propose that the most immediate obstacle to improving understanding lies in the absence of a widely applicable set of metrics for assessing transferability, and that encouraging the development of models grounded in well-established mech...

show abstract

A practical overview of transferability in species distribution modeling

Cited by 65 publications

References 107 publications

ENETwild modelling of wild boar distribution and abundance: update of occurrence and hunting data‐based models

ENETwild modelling of wild boar distribution and abundance: update of occurrence and hunting data‐based models

Model complexity affects species distribution projections under climate change

Outstanding Challenges in the Transferability of Ecological Models

Contact Info

Product

Resources

About