#STEPS TO PREVENT PSEUDOREPLICATION

library(dplyr)
library(SDMTune)
library(sp)
library(rgdal)

# FINDING CLUSTERS OF POINTS BASED ON A DISTANCE RULE 
# Having coordinate data of the observation points
xy<-SpatialPointsDataFrame(coords = data [,2:3], # indicate the columns with information for longitude and latitude
                           data=data.frame(id=data$id), #data to keep from the original dataframe
                           proj4string=crs("+init=EPSG:32738") # Project the spatial object in the right SRC before calculating distances
                           )
chc <- hclust(dist(data.frame(rownames=rownames(xy@data), x=coordinates(xy)[,1],
              y=coordinates(xy)[,2])))
# Distance with a 15km threshold  
chc<- cutree(chc, h=15000)
# Join results to meuse sp points
xy@data <- data.frame(xy@data, Clust=chc)
puntos<-data.frame(xy@data)# get a data frame with coordinates and cluster identification

##FUNCTIONS TO ITERATE THE PROCESS OF RANDOM SELECTION OF POINTS WITHIN A CLUSTER AND ANALYSIS
#preparation of different sets of combinations of points from each buffer
#puntos is the data frame with the coordinates and cluster id
#iter the number of combinations to be generated
pasteiterations<-function (puntos,iter){
result<-puntos %>% 
  group_by(Clust) %>% 
  slice_sample(n=1)
  itera<-2
repeat {
  res<-puntos %>% 
    group_by(Clust) %>% 
    slice_sample(n=1)
  result<-rbind(result,res)
  itera<-itera+1
  if (itera==iter+1) break
  
}
  
result$iteration<-rep(seq(1,iter),each=83)  
result
  
}

#RUNNING MAXENT
#presences come from the data frame of random combinations
#backgrounds is the background data used for MaxEnt
#rasters are the layers for MaxEnt
#iter the number of iterations = number of columns with random combinations generated in the previous step

multianalysis<-function (presences,backgrounds,rasters,iter){
  results<-list()
  itera<-1
  repeat {
    pres<-filter(presences,iteration==itera)
    data_T <- prepareSWD(species = "Bushpig", p = pres[,2:3], 
                           a = backgrounds,
                           env = rasters)
    folds <- randomFolds(data_T, k = 5, only_presence = TRUE, seed = 50)
    Fmodel<- train("Maxent", data = data_T, folds = folds, fc="lq",reg=0.4)
    predictions<-predict(Fmodel, data = rasters, type = "cloglog")
    
    results[itera]<-predictions
    itera<-itera+1
    if (itera==iter+1) break

  }
  
  results
  
}