Simulation Study

This repository contains the all the scripts to simulate the data, the model fitting to the simulated data and the Stan code to implement the simulation study in the manuscript “<i>High response diversity and conspecific density-dependence, not species interactions, drive dynamics of coral reef fish communities</i>”. There are 20 files in total. We ran the model fits on the HPC at James Cook University.<b>Simulated data</b><b>Sim_data_cov.R</b>: Simulates the data from the <i>Correlated responses to environmental fluctuations</i> scenario in appendix 6. It stores the simulated data in this scenario as <b>simDat_cov.RData</b>.<b>Sim_data_full.R</b>: Simulates the data from the <i>Both</i> scenario in appendix 6. It stores the simulated data in this scenario as <b>simDat_full.RData</b>.<b>Sim_data_inter.R</b>: Simulates the data from the <i>Between species interactions</i> scenario in appendix 6. It stores the simulated data in this scenario as <b>simDat_inter.RData</b>.<b>Sim_data_simple.R</b>: Simulates the data from the <i>Baseline</i> scenario in appendix 6. It stores the simulated data in this scenario as <b>simDat_simple.RData</b>.<b>Model fitting to simulated data</b><b>fit_covD2.R</b>, <b>fit_covD4.R</b>, <b>fit_covD6.R</b>, <b>fit_covD8.R</b>, <b>fit_covD10.R</b> and <b>fit_covD12.R</b> fit the model with 2, 4,6, 8, 10 and 12 latent variables, respectively, to the simulated data in the <i>Correlated responses to environmental fluctuations</i> scenario.<b>fit_full.R</b> fit the model to the simulated data in the <i>Both</i> scenario.<b>fit_inter_p0_10.R</b>, <b>fit_inter_p0_50.R</b> and <b>fit_inter_p0_100.R</b> fit the model with p0 (prior guess in the number of non-zero interspecific interactions) equals to 10, 50 and 100, respectively, to the simulated data in the <i>Between species interactions</i> scenario.<b>fit_simple.R</b> fit the model to the simulated data in the <i>Baseline</i> scenario.<b>Stan model</b><b>MARPLN_RHS_LV.stan</b>: Stan script for the multivariate autoregressive Poissson-Lognormal model with the regularised horseshoe prior and latent variables. This model does not include the Pomacentrid parameterization as the simulated species abundances do not differ in sampling area as in the LTMP data in the main text.<br>