###CREATE PP data set
#Download data file from https://doi.org/10.5066/P94GV3LC; be sure DWS file is saved as a .csv AND DELETE THE TOP TWO ROWS OF METADATA IN THE .CSV  
#set working directory to location with the DWS survey data file; 
#setwd("~/Dropbox/Dissertation/Manuscripts/Regional/RegionalMsProject/Ecosphere/Data Files")
PPdat <- read.csv("Wetland_Bird_Study_1995to1997_Dataset.csv")
head(PPdat)

PPdat <- PPdat[PPdat$pWOODL < 20,]
PPdat <- PPdat[PPdat$pSHRUB < 20,]

#Wetland area

PPdat$surveyArea <- PPdat$SIZE
PPdat$r <- sqrt((PPdat$surveyArea*10000)/pi)
PPdat$R <- PPdat$r + 400
PPdat$buffArea <- pi*(PPdat$R^2 - PPdat$r^2)/10000
PPdat$area <- (PPdat$buffArea*PPdat$pWETLA/100) + PPdat$surveyArea
PPdat$YEAR

PPdat2 <- PPdat


PPDWMPdata <- data.frame(key = PPdat2$KEY, site = PPdat2$SITEID, year = PPdat2$YEAR, 
                         OW = PPdat2$pOW, SM = PPdat2$pSM, EV = PPdat2$pEV, WM = PPdat2$pWM,
                         localSum = PPdat2$COVERSUM, 
                         localDiversity = PPdat2$COVERDIVERSITY,
                         Grass400 = PPdat2$pGRASS + PPdat2$pHAYLA,
                         Forest400 = PPdat2$pWOODL,
                         Totwet400 = PPdat2$pWETLA,
                         Crop400 = PPdat2$pCROPL,
                         nLandSum = rep(NA),
                         naturalLandDiversity = rep(NA),
                         class = PPdat2$CLASS,
                         size = PPdat2$surveyArea,
                         area = PPdat2$area,
                         PBGR = PPdat2$PBGR, AMBI = PPdat2$AMBI,
                         VIRA = PPdat2$VIRA, SORA = PPdat2$SORA)


#turn into long form
spp <- c("PBGR", "AMBI", "VIRA", "SORA")
wantedCols <- names(PPDWMPdata)[1:18]

PPlong <- data.frame()
for(i in spp){
  sppCols <- c(wantedCols, i)
  sppDf <- PPDWMPdata[sppCols]
  names(sppDf)[19] <- "count"
  sppDf$species <- rep(i)
  sppDf <- sppDf[,c(1:18, 20, 19)]
  PPlong <- rbind(PPlong, sppDf)
}

#Add in natural land and natural land diversity index
for(i in 1:nrow(PPlong)){
  PPlong$nLandSum[i] = sum(PPlong$Grass400[i], PPlong$Forest400[i], PPlong$Totwet400[i])
  PPlong$naturalLandDiversity[i] <- 1/((((PPlong$Grass400[i]/PPlong$nLandSum[i])*100)^2 + 
                                      ((PPlong$Forest400[i]/PPlong$nLandSum[i])*100)^2 +
                                      ((PPlong$Totwet400[i]/PPlong$nLandSum[i])*100)^2)/10000)
  
}

PPlong$data <- rep("PPDWMP")


#Read in MMP & CWMP DATASET
glDat <- read.csv("GL_Data.csv")

#Combine datasets and save data file
wetlandData <- rbind(PPlong, glDat)
write.csv(wetlandData, "wetlandData.csv", quote=T, row.names=F)