#--------------------------------------------------------------------------------------------------------------------- # DATA LOAD rm(list=ls()) xdata=read.csv(file="Data_Chimpanzee_Behavioural_Diversity.txt", header=T,sep=";", comment.char="", as.is=T, quote="") xdata.ori=xdata #---------------------------------------------------------------------------------------------------------------------- # DATA PREPARATION xdata.ori$Pan_tr_subspecies=relevel(as.factor(xdata.ori$Pan_tr_subspecies), "verus") xdata=subset(xdata.ori, xdata.ori$obs.pos==1) xdata$z.sq.refcdist=scale(sqrt(xdata$refcdist)) xdata$z.ObsM=scale(xdata$ObsM) xdata$z.footprint=scale(xdata$footprint) #---------------------------------------------------------------------------------------------------------------------- # LOAD PACKAGE & SETTING PRIORS library(brms) options (mc.cores=parallel::detectCores ()) prior_default = c(set_prior("normal(0,1)", class = "lscale"), set_prior("normal(0,1)", class = "sdgp") ) prior_weak = c(set_prior("normal(0,1)", class = "lscale"), set_prior("normal(0,1)", class = "sdgp"), set_prior("normal(0,1)", class = "b"), set_prior("normal(0,1)", class = "b", coef = "z.sq.refcdist"), set_prior("normal(0,1)", class = "b", coef = "z.footprint"), set_prior("normal(0,1)", class = "b", coef = "z.ObsM") ) prior_wide = c(set_prior("normal(0,1)", class = "lscale"), set_prior("normal(0,1)", class = "sdgp"), set_prior("normal(0,10)", class = "b"), set_prior("normal(0,10)", class = "b", coef = "z.sq.refugdist"), set_prior("normal(0,10)", class = "b", coef = "z.footprint"), set_prior("normal(0,10)", class = "b", coef = "z.obstime") ) #---------------------------------------------------------------------------------------------------------------------- # RUN MODEL RES.BRM.B.all=brm(obs.yn ~ z.sq.refcdist + z.footprint + z.ObsM + Pan_tr_subspecies + gp(long, lat, gr=T) #+(1|commu) +(1+z.sq.refcdist + z.footprint + z.ObsM|Site) +(1+z.sq.refcdist + z.footprint + z.ObsM + Pan_tr_subspecies|Behavior) , data=xdata, family=bernoulli, prior = prior_weak, sample_prior = TRUE,warmup = 1000,iter = 2000, chains = 4,control = list(adapt_delta = 0.99, max_treedepth = 16)) #---------------------------------------------------------------------------------------------------------------------- # MODEL RESULT # > summary(RES.BRM.B.all) # Family: bernoulli # Links: mu = logit # Formula: obs.yn ~ z.sq.refcdist + z.footprint + z.ObsM + Pan_tr_subspecies + gp(long, lat, gr = T) + (1 + z.sq.refcdist + z.footprint + z.ObsM | Site) + (1 + z.sq.refcdist + z.footprint + z.ObsM + Pan_tr_subspecies | Behavior) # Data: xdata (Number of observations: 4402) # Samples: 4 chains, each with iter = 2000; warmup = 1000; thin = 1; # total post-warmup samples = 4000 # Gaussian Process Terms: # Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat # sdgp(gplonglat) 0.71 0.17 0.37 1.02 631 1.01 # lscale(gplonglat) 0.00 0.00 0.00 0.00 1383 1.00 # Group-Level Effects: # ~Behavior (Number of levels: 31) # Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat # sd(Intercept) 1.92 0.36 1.31 2.73 1411 1.00 # sd(z.sq.refcdist) 0.76 0.17 0.47 1.14 1282 1.00 # sd(z.footprint) 0.13 0.10 0.00 0.36 2499 1.00 # sd(z.ObsM) 0.80 0.17 0.51 1.20 2447 1.00 # sd(Pan_tr_subspeciesellioti) 2.16 0.74 0.98 3.84 2239 1.00 # sd(Pan_tr_subspeciesschweinfurthii) 2.17 0.49 1.36 3.27 2794 1.00 # sd(Pan_tr_subspeciestroglodytes) 2.98 0.63 1.98 4.42 2560 1.00 # cor(Intercept,z.sq.refcdist) -0.17 0.22 -0.57 0.27 2469 1.00 # cor(Intercept,z.footprint) 0.09 0.34 -0.59 0.70 6318 1.00 # cor(z.sq.refcdist,z.footprint) 0.18 0.36 -0.54 0.79 6383 1.00 # cor(Intercept,z.ObsM) 0.03 0.20 -0.35 0.41 3268 1.00 # cor(z.sq.refcdist,z.ObsM) 0.04 0.22 -0.39 0.47 2292 1.00 # cor(z.footprint,z.ObsM) -0.03 0.34 -0.66 0.64 420 1.01 # cor(Intercept,Pan_tr_subspeciesellioti) 0.04 0.25 -0.44 0.51 4140 1.00 # cor(z.sq.refcdist,Pan_tr_subspeciesellioti) 0.32 0.26 -0.23 0.78 2826 1.00 # cor(z.footprint,Pan_tr_subspeciesellioti) 0.15 0.34 -0.54 0.76 933 1.00 # cor(z.ObsM,Pan_tr_subspeciesellioti) 0.11 0.25 -0.37 0.57 3583 1.00 # cor(Intercept,Pan_tr_subspeciesschweinfurthii) -0.09 0.20 -0.47 0.30 3116 1.00 # cor(z.sq.refcdist,Pan_tr_subspeciesschweinfurthii) 0.21 0.22 -0.24 0.63 2560 1.00 # cor(z.footprint,Pan_tr_subspeciesschweinfurthii) 0.10 0.35 -0.60 0.71 491 1.00 # cor(z.ObsM,Pan_tr_subspeciesschweinfurthii) 0.20 0.22 -0.24 0.60 2810 1.00 # cor(Pan_tr_subspeciesellioti,Pan_tr_subspeciesschweinfurthii) 0.40 0.23 -0.08 0.78 2151 1.00 # cor(Intercept,Pan_tr_subspeciestroglodytes) -0.11 0.19 -0.47 0.28 3792 1.00 # cor(z.sq.refcdist,Pan_tr_subspeciestroglodytes) 0.15 0.22 -0.28 0.56 3031 1.00 # cor(z.footprint,Pan_tr_subspeciestroglodytes) 0.13 0.34 -0.54 0.71 417 1.01 # cor(z.ObsM,Pan_tr_subspeciestroglodytes) 0.35 0.20 -0.06 0.69 3397 1.00 # cor(Pan_tr_subspeciesellioti,Pan_tr_subspeciestroglodytes) 0.53 0.19 0.10 0.84 2786 1.00 # cor(Pan_tr_subspeciesschweinfurthii,Pan_tr_subspeciestroglodytes) 0.35 0.20 -0.06 0.69 3635 1.00 # ~Site (Number of levels: 107) # Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat # sd(Intercept) 0.45 0.27 0.02 1.00 540 1.00 # sd(z.sq.refcdist) 0.64 0.27 0.07 1.13 557 1.00 # sd(z.footprint) 0.19 0.15 0.01 0.58 1426 1.00 # sd(z.ObsM) 0.83 0.28 0.37 1.44 1205 1.00 # cor(Intercept,z.sq.refcdist) 0.09 0.40 -0.68 0.81 579 1.01 # cor(Intercept,z.footprint) 0.06 0.45 -0.80 0.85 3880 1.00 # cor(z.sq.refcdist,z.footprint) 0.05 0.45 -0.79 0.83 3842 1.00 # cor(Intercept,z.ObsM) 0.08 0.43 -0.75 0.82 990 1.00 # cor(z.sq.refcdist,z.ObsM) -0.19 0.39 -0.84 0.65 950 1.01 # cor(z.footprint,z.ObsM) 0.01 0.44 -0.80 0.79 1472 1.00 # Population-Level Effects: # Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat # Intercept -4.24 0.47 -5.18 -3.35 1640 1.00 # z.sq.refcdist 0.52 0.23 0.07 0.98 2909 1.00 # z.footprint -0.25 0.16 -0.58 0.04 3359 1.00 # z.ObsM 0.87 0.30 0.29 1.48 3539 1.00 # Pan_tr_subspeciesellioti 0.12 0.69 -1.29 1.41 4516 1.00 # Pan_tr_subspeciesschweinfurthii -0.11 0.51 -1.13 0.87 3607 1.00 # Pan_tr_subspeciestroglodytes 0.15 0.64 -1.16 1.40 4044 1.00 # Samples were drawn using sampling(NUTS). For each parameter, Eff.Sample # is a crude measure of effective sample size, and Rhat is the potential # scale reduction factor on split chains (at convergence, Rhat = 1). # > # > ##################################################################################################################################