### R scripts for importing and curating GBIF records
### R. Castilho || CCMAR || 2015-03-15



### GOALS
# 1. Import GBIF records
# 2. Map all the points
# 3. Curate reference points
# 4. Make final map


############
### CODE ###
############

#######################################
### STEP 1 PREPARATION STEPS
# see xls example file http://rcastilho.pt/BEMO/GBIF_data_example.xlsx
# install packages
install.packages(c('xlsx', 'maptools', 'graphics')) 
# load packages 
library(xlsx)
library(maptools)
library(graphics)

#######################################
### STEP 2 LOAD SPECIES DATA
# set working directory... change this to the directory in which the species' data is stored and where you want to save output
setwd("D:/UTILIZADORES/user-en/Desktop/XXX")
# load species' records
location.data<-read.xlsx("XXX.xlsx",1)

codes<-location.data$code
y<-location.data$lat
x<-location.data$long


#######################################
# REMOVING DUPLICATES
# which records are duplicates?
dups <- !duplicated(location.data[, 2:3])
dups
# remove duplicates
location.data  <-  location.data[dups, ]
location.data[,1:3]

#######################################
# reducing the geographical area to decrease computation time
# here are the minimum and maximum lat-long values of the data
#HOW TO DO: if max(x)<0, max(x)=max(x)-5 otherwise max(x)=max(x)+5
xmin=min(x)-5
xmax=max(x)+5
ymin=min(y)-5
ymax=max(y)+5

#######################################
# PLOT MAP
data(wrld_simpl)
plot(wrld_simpl, xlim=c(xmin,xmax), ylim=c(ymin,ymax), axes=TRUE, col="light grey")
# plot points
points(x, y, col='red', pch=20, cex=0.5) 
# plot points again to add a border, for better visibility
points(x, y, col='black', cex=0.75)

# put labels to simplify id of odd data
text(x, y, labs = codes, offset=0.5, cex=1)

# removing odd points
# e.g.1836
location.data =location.data[-1836,]
location.data =location.data[-XX,]
location.data =location.data[-XX]
location.data =location.data[-XX,]
location.data =location.data[-XX,]
location.data =location.data[-XX,]
location.data =location.data[-XX,]
location.data =location.data[-XX,]
location.data =location.data[-XX,]
location.data =location.data[-XX,]
location.data =location.data[-XX,]
location.data =location.data[-XX,]

# repeat map
y<-location.data$lat
x<-location.data$long
xmin=min(x)-5
xmax=max(x)+5
ymin=min(y)-5
ymax=max(y)+5
data(wrld_simpl)
plot(wrld_simpl, xlim=c(xmin,xmax), ylim=c(ymin,ymax), axes=TRUE, col="light grey")
# plot points
points(x, y, col='blue', pch=20, cex=0.5) 
# plot points again to add a border, for better visibility
points(x, y, col='black', cex=0.5)