landscape-sim

This repository contains files to run movement simulations and analyze resulting data for the PNAS 2018 manuscript: L.A. White, J.D. Forester & M.E. Craft. "Disease outbreak thresholds emerge from interactions between movement behavior, landscape structure, and epidemiology"

www.pnas.org/cgi/doi/10.1073/pnas.1801383115

Scripts:

1. sim_parallel_functions.R- contains IBM_loop_function and sub functions needed to run simulations
2. move_sim1.R: Sample script to run IBM_loop_function in parallel via clusters
3. Manuscript_figures_2017_12_12.Rmd- Analyzes IBMsummary.csv simulation results to produce figures for manuscript
4. RandomForestParallelLogit.R- random forest analysis to test- is an outbreak successful?
5. RandomForestParallelLogitDur.R- given a successful outbreak- what determines outbreak duration? RandomForestParallelLogitPrev.R- given a successful outbreak- what determines maximum prevalence?
6. SimpleSIR.Rmd- compare spatially-explicit simulations to comparable stochastic model that assumes homogeneous mixing (produces Figure S7)
7. PartyRFLogit.R- random forest analysis (using cforest function from party package) to test- is an outbreak successful?
8. PartyRFLogitPrev.R-random forest analysis (using cforest function from party package) to test- is an outbreak successful?
9. PartyRFLogitDur.R- random forest analysis (using cforest function from party package) to test- what determines outbreak duration?

Data/results:

1. IBMsummary.zip- contains IBMsummary.csv with all simulation results; headers: X= parameter set/combination; k= size of landscape (2^k+1); density= conspecific density (0.25 or 0.5); rec_rate=recovery rate; p=proportion available habitat; H=Hurst exponent; beta1 =strength of selection for resources; beta2= strength of selection for conspecifics; beta3= squared term in RSF for strength of selection of conspecifics; percep= perceptual range (1,2, or 3 cells out from current cell); duration= duration of outbreak; max_I= maximum number of individuals infected at any given time during outbreak; max_prev= maximum prevalence; betas= character string of betas used for parameter set; pbyH= string of values used for p and H to create landscape structure
2. imp_logit.csv and imp_logitSD.csv- variable importance and standard deviation (SD) results from RandomForestParallelLogit.R
3. logit_imp_dur.csv and imp_logitdurSD.csv- variable importance and standard deviation (SD) results from RandomForestParallelLogitDur.R
4. logit_imp_maxprev.csv and imp_logitprevSD.csv- variable importance and standard deviation (SD) results from RandomForestParallelLogitPrev.R
5. SIRdata_merged.csv- simplistic, stochastic, homogeneous SIR model results produced from SimpleSIR.Rmd
6. partyRF_logit1000.csv- variable importance reuslts from PartyRFLogit.R using 1,000 trees
7. partyRF_logitdur1000.csv - variable importance reuslts from PartyRFLogitDur.R using 1,000 trees
8. partyRF_logitprev1000.csv - variable importance reuslts from PartyRFLogitPrev.R using 1,000 trees