ABC random forests for Bayesian parameter inference

Raynal, Louis; Marin, Jean‐Michel; Pudlo, Pierre; Ribatet, Mathieu; Robert, Christian P.; Estoup, Arnaud

doi:10.1093/bioinformatics/bty867

Cited by 174 publications

(251 citation statements)

References 51 publications

Supporting

Mentioning

251

Contrasting

Order By: Relevance

“…We believe that we are the first to employ this technique for dimension reduction in ABC rejection sampling, although Raynal et al. () used it to directly estimate model parameter values from the parameter sample. Whilst useful, their approach does not approximate the parameter posterior distribution.…”

Section: Discussionmentioning

confidence: 99%

Calibrating an individual‐based movement model to predict functional connectivity for little owls

Hauenstein

Fattebert

Grüebler

et al. 2019

Ecological Applications

View full text Add to dashboard Cite

Dispersal is crucial for population viability and thus a popular target for conservation measures. However, the ability of individuals to move between habitat patches is notoriously difficult to estimate. One solution is to quantify functional connectivity via realistic individual‐based movement models. Such simulation models, however, are difficult to build and even more difficult to parameterize. Here, we use the example of natal little owl (Athene noctua) dispersal to develop a new analysis chain for the calibration of individual‐based dispersal models using a hybrid of statistical parameter estimation and Approximate Bayesian Computation (ABC). Specifically, we use locations of 126 radio‐tracked juveniles to first estimate habitat utilization by generalized additive models (GAMs) and the biased random bridges (BRB) method. We then include the estimated parameters in a spatially explicit individual‐based model (IBM) of little owl dispersal and calibrate further movement parameters using ABC. To derive efficient summary statistics, we use a new dimension reduction method based on random forest (RF) regression. Finally, we use the calibrated IBM to predict the dispersal potential of little owls from local populations in southwestern Germany to suitable habitat patches in northern Switzerland. We show that pre‐calibrating habitat preference parameters while inferring movement behavioral parameters via ABC is a computationally efficient solution to obtain a plausible IBM parameterization. We also find that dimension reduction via RF regression outperforms the widely used least squares regression, which we applied as a benchmark approach. Estimated movement parameters for the individuals reveal plausible inter‐individual and inter‐sexual differences in movement behavior during natal dispersal. In agreement with a sex‐biased dispersal distance in little owls, females show longer individual flights and higher directional persistence. Simulations from the fitted model indicate that a (re)colonization of northern Switzerland is generally possible, albeit restricted. We conclude that the presented analysis chain is a sensible work‐flow to assess dispersal connectivity across species and ecosystems. It embraces species‐ and individual‐specific behavioral responses to the landscape and allows likelihood‐based calibration, despite an irregular sampling design. Our study highlights existing, yet narrow dispersal corridors, which may require enhancements to facilitate a recolonization of little owl habitat patches in northern Switzerland.

show abstract

Section: Discussionmentioning

confidence: 99%

Calibrating an individual‐based movement model to predict functional connectivity for little owls

Hauenstein

Fattebert

Grüebler

et al. 2019

Ecological Applications

View full text Add to dashboard Cite

show abstract

“…The resulting observed values of summary statistics were compared to 1,000 simulated values from data sets for which parameters were drawn from their posterior distribution (posterior model checking). Finally, posterior distributions of parameters were estimated for the final selected scenario using 10,000 simulated data sets under a regression by random forest methodology (Raynal et al , arXiv), with classification forests of 1,000 trees.…”

Section: Methodsmentioning

confidence: 99%

Deciphering the worldwide invasion of the Asian long‐horned beetle: A recurrent invasion process from the native area together with a bridgehead effect

Javal¹,

Lombaert

Tsykun

et al. 2019

Molecular Ecology

View full text Add to dashboard Cite

Retracing introduction routes is crucial for understanding the evolutionary processes involved in an invasion, as well as for highlighting the invasion history of a species at the global scale. The Asian long‐horned beetle (ALB) Anoplophora glabripennis is a xylophagous pest native to Asia and invasive in North America and Europe. It is responsible for severe losses of urban trees, in both its native and invaded ranges. Based on historical and genetic data, several hypotheses have been formulated concerning its invasion history, including the possibility of multiple introductions from the native zone and secondary dispersal within the invaded areas, but none have been formally tested. In this study, we characterized the genetic structure of ALB in both its native and invaded ranges using microsatellites. In order to test different invasion scenarios, we used an approximate Bayesian “random forest” algorithm together with traditional population genetics approaches. The strong population differentiation observed in the native area was not geographically structured, suggesting complex migration events that were probably human‐mediated. Both native and invasive populations had low genetic diversity, but this characteristic did not prevent the success of the ALB invasions. Our results highlight the complexity of invasion pathways for insect pests. Specifically, our findings indicate that invasive species might be repeatedly introduced from their native range, and they emphasize the importance of multiple, human‐mediated introductions in successful invasions. Finally, our results demonstrate that invasive species can spread across continents following a bridgehead path, in which an invasive population may have acted as a source for another invasion.

show abstract

“…We used the newly developed ABC method based on a machine learning tool named “random forest” (ABC‐RF) to perform model choice and parameter estimation (Pudlo et al, ; Raynal et al, ). This approach allows disentangling complex demographic models (Pudlo et al, ), by comparing groups of scenarios with a specific type of evolutionary event to other groups with different types of evolutionary events (instead of considering all scenarios separately; Estoup, Raynal, Verdu, & Marin, ), in what we will hereafter call “ABC rounds.” We used sequential rounds to compare a group of scenarios with gene flow to a group of scenarios without gene flow, a group of scenarios assuming independent domestication events from different wild lineages to a group of scenarios assuming repeated domestication events from the same wild lineage, and a group of scenarios assuming bottlenecks to a group of scenarios assuming no bottleneck during domestication.…”

Section: Methodsmentioning

confidence: 99%

The complex evolutionary history of apricots: Species divergence, gene flow and multiple domestication events

et al. 2019

View full text Add to dashboard Cite

Domestication is an excellent model to study diversification and this evolutionary process can be different in perennial plants, such as fruit trees, compared to annual crops. Here, we inferred the history of wild apricot species divergence and of apricot domestication history across Eurasia, with a special focus on Central and Eastern Asia, based on microsatellite markers and approximate Bayesian computation. We significantly extended our previous sampling of apricots in Europe and Central Asia towards Eastern Asia, resulting in a total sample of 271 cultivated samples and 306 wild apricots across Eurasia, mainly Prunus armeniaca and Prunus sibirica, with some Prunus mume and Prunus mandshurica. We recovered wild Chinese species as genetically differentiated clusters, with P. sibirica being divided into two clusters, one possibly resulting from hybridization with P. armeniaca. Central Asia also appeared as a diversification centre of wild apricots. We further revealed at least three domestication events, without bottlenecks, that gave rise to European, Southern Central Asian and Chinese cultivated apricots, with ancient gene flow among them. The domestication event in China possibly resulted from ancient hybridization between wild populations from Central and Eastern Asia. We also detected extensive footprints of recent admixture in all groups of cultivated apricots. Our results thus show that apricot is an excellent model for studying speciation and domestication in long‐lived perennial fruit trees.

show abstract

ABC random forests for Bayesian parameter inference

Abstract: Please refer to published version for the most recent bibliographic citation information. If a published version is known of, the repository item page linked to above, will contain details on accessing it.

Cited by 174 publications

References 51 publications

Calibrating an individual‐based movement model to predict functional connectivity for little owls

Calibrating an individual‐based movement model to predict functional connectivity for little owls

Deciphering the worldwide invasion of the Asian long‐horned beetle: A recurrent invasion process from the native area together with a bridgehead effect

The complex evolutionary history of apricots: Species divergence, gene flow and multiple domestication events

Contact Info

Product

Resources

About