Chapter 5 Computational methods

Now, let’s come to the more practical part of our Beginner’s Guide. Here, you will get an overview on the methods that we use in our working group.

5.1 Sensitivity Analyses

Methods that determine how target variables are affected based on changes in other variables known as input variables.

5.1.1 Overview

General broad overview in Sensitivity Analysis in Practice : A Guide to Assessing Scientific Models (Saltelli et al. 2002)

Comparison of different sensitivity analysis methods: A performance comparison of sensitivity analysis methods for building energy models Idea: Run the sensitivity analysis not on the data itself but on a comparison between data and model (as you would use in the validation) -> This is the measure to which you want to know the sensitivity

5.1.2 Local

= One-factor-at-a-time (OAT)

TODO Anne

5.1.3 Global

Morris screening and enhancement as used in the R package sensitivity (Campolongo 2007).

TODO Anne

5.2 Virtual ecologist

TODO

5.3 Parameter optimization

5.3.1 Direct methods

Differential evolution optimization Differential Evolution with DEoptim (R package) How does differential evolution work?

TODO Anne

5.3.2 Bayesian Calibration

florianhartig/LearningBayes: An introduction to Bayesian statistics Bayesian Tools - General-Purpose MCMC and SMC Samplers and Tools for Bayesian Statistics Inverse modeling https://arxiv.org/ftp/arxiv/papers/2007/2007.15580.pdf

5.3.3 Emulatoren

Approximieren das Modell selbst “modelling a model” / “meta-model”

TODO

5.3.5 Tips and Tricks

For all methods, first, run an optimization on a dataset that you produced with the model. That way, you can find out whether the optimization algorithm can find the parameters with which you modelled the dataset. This can help to verify that your optimization method works.

5.4 Machine Learning

If you are interested in machine learning, here is a course by our colleague Florian Hartig from the University of Regensburg. This course is held in R.

5.5 Phylogenetic Analyses

TODO

5.6 Functional Analyses

Wright et al (2005) provides a good basis for ‘trusting’ trait-trait associations above 5-fold variation and/or n=20 spp, within a study (Wright et al. 2005). Although Wright et al. show trait correlations with as few as 4 species.

For multivariate analyses:

  • MANOVA (because traits are not independent from one another) and canonical discriminant analyses (CDA) =Y check R package candisc

  • trait network analyses (R package igraph - Csardi and Nepusz, 2006)

    • Mantel test for comparing trait correlations across groups (e.g. taxonomic, trophic or guild groups)

  • PCA (useful for visualization as well)

5.7 SDMs

ToDo: WHAT IS A SDM?

TODO: organize better and make references readable

Tutorials:

Further:

Earlier models:

Packages:

For Maxent type of models:

For ensemble models:

Useful literature:

  • Review about SDMs (Hao et al. 2019)

  • ODMAP: A standard protocol for reporting SDMs (Zurell et al. 2020)

  • Correcting sampling bias in MAxent (Kramer-Schadt et al. 2013)

References

Hao, Tianxiao, Jane Elith, Gurutzeta Guillera-Arroita, and José J. Lahoz-Monfort. 2019. “A Review of Evidence About Use and Performance of Species Distribution Modelling Ensembles Like Biomod.” Edited by Josep Serra-Diaz. Diversity and Distributions 25 (5): 839–52. https://doi.org/10.1111/ddi.12892.

Kramer-Schadt, Stephanie, Jürgen Niedballa, John D. Pilgrim, Boris Schröder, Jana Lindenborn, Vanessa Reinfelder, Milena Stillfried, et al. 2013. “The Importance of Correcting for Sampling Bias in Maxent Species Distribution Models.” Edited by Mark Robertson. Diversity and Distributions 19 (11): 1366–79. https://doi.org/10.1111/ddi.12096.

Leroy, Boris, Christine N. Meynard, Céline Bellard, and Franck Courchamp. 2015. “Virtualspecies, an R Package to Generate Virtual Species Distributions.” Ecography 39 (6): 599–607. https://doi.org/10.1111/ecog.01388.

Saltelli, Andrea, Stefano Tarantola, Francesca Campolongo, and Marco Ratto. 2002. “Sensitivity Analysis in Practice,” February. https://doi.org/10.1002/0470870958.

Wright, Ian J., Peter B. Reich, Johannes H. C. Cornelissen, Daniel S. Falster, Eric Garnier, Kouki Hikosaka, Byron B. Lamont, et al. 2005. “Assessing the Generality of Global Leaf Trait Relationships.” New Phytologist 166 (2): 485–96. https://doi.org/10.1111/j.1469-8137.2005.01349.x.

Zurell, Damaris, Janet Franklin, Christian König, Phil J. Bouchet, Carsten F. Dormann, Jane Elith, Guillermo Fandos, et al. 2020. “A Standard Protocol for Reporting Species Distribution Models.” Ecography 43 (9): 1261–77. https://doi.org/10.1111/ecog.04960.