04-Results.Rmd

---
output:
  word_document: default
  pdf_document: default
  html_document: default
---

```{r, echo = FALSE}
editMetric <- function(metriclist, colnumber){

      ee <- lapply(lapply(metriclist, summary), function(x) x[,colnumber]) # Focus on column
      ee <- gsub("Min.   :|1st Qu.:|Median :|Mean   :|3rd Qu.:|Max.   :", "", ee) # Erase strings from summary output 
      ee <- regmatches(ee,
                       gregexpr("[-+]?[0-9]*\\.?[0-9]+([eE][-+]?[0-9]+)?",ee)) # Match floating point with regex (http://www.regular-expressions.info/floatingpoint.html)
      ee <- lapply(ee, as.numeric) # String to numbers 
      ee1 <- c()
        for(i in 1:6){
          ee1[[i]] <- unlist(ee)[seq(i,length(unlist(ee)), by= 6)] # Sequence data from summary output
        }
      names(ee1) <- c("Min", "firstQ", "Median", "Mean", "thirdQ", "Max")
      
      ee1$distances <- unlist(as.numeric(regmatches(names(metric),
                                                    gregexpr("[-+]?([0-9]*[0-9]+|[0-9]+)",
                                                             names(metriclist))))) # Add values of distances based on named files

      return(ee1)
     
}

```

### Raphia taedigera seed dispersal 


```{r, echo = FALSE,  out.width='400px', fig.align='center', fig.cap="Distribution of Raphia dominated palm swamps (Yolillales) in Costa Rica and Nicaragua.  Data comes from Serrano - Sandi et al. 2013. Interactive version in: https://code.earthengine.google.com/49bfde3347f6a527d733d20c1dae4c2d"}
knitr::include_graphics("www/images/distribution.png")
```

```{r, echo=FALSE, fig.align='center', fig.cap="Ecological observations gathered from literature. Bars represent the number of observations for a determined Ecological class or from a determined source. Gray bars represent the total number of observations, colored bars represent only those specific to Tapirus bairdii"}

obs <- read.csv("www/observation/Observations.csv", header = T, stringsAsFactors = T)

col = c("#777acd",
"#ad963d",
"#c55a9f",
"#5ba965",
"#ca5e4a")

col2 = c("#6bb84c",
"#b460bd",
"#c7aa3f",
"#697ecd",
"#cf5f40",
"#4fb9a1",
"#c8567a",
"#64843f",
"#ac7d41",
"darkgreen")

par(las = 1, mfrow = c(2,2), oma = c(0,0,0,0), mar = c(0.5,4,1,3))
barplot(table(obs$VariableClass), 
        xaxt="n", 
        xlab = "Classes",
        ylab = "Observations")
barplot(table(obs$VariableClass[obs$Species == "Tapirus bairdii"]), 
        xaxt="n", 
        xlab = "Classes",
        ylab = "Observations", 
        add = T,
        col = col)
plot(0,0, xaxt = "n", frame.plot = F,  xlab = "", ylab= "", yaxt = "n")
legend("top", levels(obs$VariableClass), fill = col, title = "Ecology classes")

barplot(table(obs$Source), 
        xaxt="n", 
        xlab = "Classes",
        ylab = "Observations")
barplot(table(obs$Source[obs$Species == "Tapirus bairdii"]), 
        xaxt="n", 
        xlab = "Classes",
        ylab = "Observations", 
        add = T,
        col = col2)
plot(0,0, xaxt = "n", frame.plot = F,  xlab = "", ylab= "", yaxt = "n")
legend("top", levels(obs$Source), fill = col2, title = "Source")


```

# Probability of connectivity


```{r, echo=FALSE}
# Load data from CONEFOR (PC deltas)
coneforMetrics <- list.files("www/data/CONEFORmetrics/") # Files in data folder 
coneforMetrics <- paste("www/data/CONEFORmetrics/", coneforMetrics, sep="") # Recreate full path 
overall <- list.files("www/data/OverallIndex/")
overall <-  paste("www/data/OverallIndex/", overall, sep="")

metric <- lapply(coneforMetrics, function(x) read.csv(x, header = T, stringsAsFactors = F)) # Load individual csv's
names(metric) <- list.files("www/data/CONEFORmetrics/") # Give appropiate names

ov.metric <- lapply(overall, function(x) read.table(x, header = F, stringsAsFactors = F))
names(ov.metric) <- list.files("www/data/OverallIndex/") # Give appropiate names

PCnum <- reshape2::melt(lapply(ov.metric, function(x) x))
PCnum$L1 <- as.numeric(gsub(".txt", "", PCnum$L1))
PCnum$variable <- NULL

  
# 
# # Max dPA value to calculate PC 
# al <- max(metric$`100m.csv`$dA) ### To Check maximun probability!!! 
# PCmetri <- lapply(metric ,function (x) sapply(x[,3], function(x) x/al^2)) # Calculate PC metrics from Conefor delta outputs
# names(PCmetri) <- as.numeric(sapply(names(PCmetri), function(x)gsub( "m.csv", "",x))) # extract the "m.csv" pattern from the names 
# PCmetri <- lapply(PCmetri, function(x) sort(x, decreasing = T)) # Sort PC (useful to apply colors in maps later)
# PCmetri <- reshape2::melt(PCmetri) # flatten list
# PCmetri$L1 <- as.numeric(PCmetri$L1) # as numeric threshold distances


dPC <- lapply(metric ,function (x) sapply(x[,3], function(x) x)) # PCnum values
names(dPC) <- as.numeric(sapply(names(dPC), function(x)gsub( "m.csv", "",x))) # extract the "m.csv" pattern from the names 
dPC <- lapply(dPC, function(x) sort(x, decreasing = T)) # Sort PC (useful to apply colors in maps later)
dPC <- reshape2::melt(dPC) # flatten list
dPC$L1 <- as.numeric(dPC$L1) # as numeric threshold distances

```

```{r, echo = FALSE}
#####Settings
colpal = c('#67001f','#b2182b','#d6604d','#f4a582','#fddbc7','#d1e5f0','#92c5de','#4393c3','#2166ac','#053061') # Set palette


names(colpal) = sort(unique(as.numeric(PCnum$L1)), decreasing = T) # name colors according to dispersal distances
#PCmetri$col <- colpal[match( PCmetri$L1,names(colpal))]
PCnum$col <- colpal[match( PCnum$L1,names(colpal))]
#PCmetri$col <- colpal[match( PCmetri$L1,names(colpal))]
dPC$col <- colpal[match( dPC$L1,names(colpal))]
colr = scales::alpha("#a7c78e",alpha = 0.3)
######
```


```{r, echo = FALSE, eval = T}
# Load shapefile with the polygons with Raphia distributions
dist <-raster::shapefile("www/data/raphiadist/patchessinglepart.shp")
# Merge the shapefile with the metrics dataset
dist@data <- as.data.frame(sapply(metric, function(x) raster::merge(dist,x, by.y ="Node", by.x = "Patches")))

# Write output as file 
# rgdal::writeOGR(dist, "www/data/raphiadist/merge", driver = "ESRI Shapefile", layer = "dist")
# merged <- rgdal::readOGR("www/data/raphiadist/merge/dist.shp", "dist")
```

```{r, echo = FALSE, results=FALSE, fig.align="center", fig.cap=" Probability of Connectivity (PCnum) for all distance thresholds evaluated. Left pane shows the distribution of PCnum values for all patches at all dispersal distances considered, PCnum has been converted to Equivalent Connected Area (ECA) (Saura 2011) to prioritize better low metric values of PCnum (Saura and Torne 2012). Right panes show the median (up) and mean (bottom) PCnum values in function of the distance thresholds. Note that the x axis has been logarithimized for all plots. Highligthed region correspond to the distances for which dispersl thresholds correspond to the maximum dispersal distances and daily movement of Tapirs reported in literature"}
split.screen(rbind(c(0.15,0.6,0.05, 0.98), c(0.6, 0.98, 0.05, 0.6), c(0.6, 0.98, 0.6, 0.98), c(0.0, 0.13, 0.1, 0.98), c(0.01,0.9,0.01,0.2)))
screen(1)
par(las = 2, mar = c(4,4,0,1), oma = c(1,1,1,1))

plot(value~jitter(L1, 0.5), dPC,
     log ="x", pch = 16, 
     ylab= "dPC = dPCIntra + dPCFlux + dPCConnector", 
     xlab = "Distance threshold", 
     col = dPC$col, cex = 1.2,
     cex.lab = 0.7, cex.axis = 0.7)

rect(1500, 0,52000,30, col = colr, border = NA)
screen(2)
par(las=2, mar = c(4,4,0,0))

plot(sqrt(value)~L1 , PCnum,
     log ="x", pch = 16,
     ylab= " ECA = sqrt(PCnum)", 
     xlab = "Distance threshold", 
     col = PCnum$col, cex = 1.2,
     cex.lab = 0.7, cex.axis = 0.7)
points(sqrt(value)~L1 , PCnum, cex = 1.2)
rect(1500, 0,52000,16E09, col = colr, border = NA)

screen(3)
par(las=2, mar = c(0,4,0,0))

plot(sapply(unique(L1), function(x) median(value[L1 == x])) ~ unique(L1) , dPC, 
      cex = 1.2,  col = unique(PCnum$col), 
     pch = 15, log = "x",
     cex.lab = 0.7, cex.axis = 0.7, 
     xaxt = "n", ylab = "median dPC")
points(sapply(unique(L1), function(x) median(value[L1 == x])) ~ unique(L1) , dPC, 
       cex = 1.2, pch = 0,
       cex.lab = 0.7, cex.axis = 0.7)

rect(1500, 0,52000,0.8, col = colr, border = NA)
screen(4)
par(mar = c(0,0,0,0), oma = c(0,1,2,0))
plot.new()
legend("center", legend = sort(unique(dPC$L1)),
       pch =16, col = rev(colpal), title = "Distance \n threshold \n (meters)", bty = "n")
```


Export figures for report: 

Hyphotesis 1:  
```{r}

png(filename = "www/ReportFilesPlots/Hyphothesis1.png",width = 1900, height = 1500, res = 250  )
par(las = 1, oma = c(3,3,3,3))

bars <- rev(c('blue','#feb24c','#f03b20'))

colr = scales::alpha(bars,alpha = 0.3)
dat <- PCnum[PCnum$V1 == "EC(PC):",]
plot(value~L1 ,dat ,
     pch = 16,log = "x",
     ylab= " EC (Equivalent Connectivity)", 
     xlab = "Maximum Dispersal Distance threshold Log(m)", 
     col = "black", cex = 1.5,
     cex.lab = 1.2, cex.axis = 0.7)

rect(13000, 0, 100, 4.466596e+08, col = colr[1], border = NA)
rect(13000, 52000,52000,5.754883e+08, col =colr[2] , border = NA)
rect(52000, 0 ,240000,10e8, col = colr[3], border = NA)
points(value~L1 , dat, cex = 1.2)
legend("bottomright",c("Home Range", "Dispersal", "Migration"), fill = bars, title = "Type of movement")

dev.off()
```



Hypothesis 2:

```{r}

png(filename = "www/ReportFilesPlots/Hyphothesis2.1.png",width = 1900, height = 1500, res = 250  )
par(las = 1, oma = c(3,3,3,3))

plot(value~jitter(L1, 0.5), dPC,
     log ="x", pch = 16, 
     ylab= "dPC = dPCIntra + dPCFlux + dPCConnector", 
     xlab = "Maximum Dispersal Distance threshold Log(m)", 
     col = dPC$col, cex = 1.2,
     cex.lab = 1.2, cex.axis = 0.7)
rect(13000, 0, 100, 30, col = colr[1], border = NA)
rect(13000, 0,52000,23, col =colr[2] , border = NA)
rect(52000, 0 ,240000,16, col = colr[3], border = NA)
points(value~jitter(L1, 0.5), dPC,
     col = dPC$col, cex = 1.2, pch = 16)
dev.off()
```


```{r}
png(filename = "www/ReportFilesPlots/Hyphothesis2.2.png",width = 1900, height = 1500, res = 250  )
par(las =1,oma = c(3,3,3,3) )
plot(sapply(unique(L1), function(x) median(value[L1 == x])) ~ unique(L1) , dPC, 
      cex = 1.5,  col = unique(PCnum$col), 
     pch = 15, log = "x",
     cex.lab = 1.2, cex.axis = 0.7, 
     ylab = "median dPC",
     xlab = "Maximum Dispersal Distance threshold Log(m)" )
rect(13000, 0, 100, 0.026, col = colr[1], border = NA)
rect(13000, 0,52000,0.043, col =colr[2] , border = NA)
rect(52000, 0 ,240000,0.05, col = colr[3], border = NA)
points(sapply(unique(L1), function(x) median(value[L1 == x])) ~ unique(L1) , dPC, 
       cex = 1.2, pch = 0,
       cex.lab = 1.2, cex.axis = 0.7)
legend("topleft",c("Home Range", "Dispersal", "Migration"), fill = bars, title = "Type of movement")
dev.off()
```



```{r}

png(filename = "www/ReportFilesPlots/Hyphothesis2.png",width = 1900, height = 1500, res = 250  )
par(las =1,oma = c(3,3,3,3) )
meds <- sapply(unique(dPC$L1) ,function(x) median(dPC$value[dPC$L1 == x]))

meds <- reshape2::melt(lapply(meds, function(x) rep(x, 732)))

dPC$med <- meds$value
dPC$rad <- dPC$value# Relative removal effect

rad <- aggregate(dPC$value, FUN = "scale", by = list(dPC$L1))$x # Mean relative removal efffect
rad <- reshape2::melt(rad)

rad$sd <- aggregate(dPC$value, FUN = "sd", by = list(dPC$L1))$x

avgrad <-  aggregate(rad$value, FUN = "median", by = list(rad$Var1))$x
maxrad <-  aggregate(rad$value, FUN = "max", by = list(rad$Var1))$x
minrad <- aggregate(rad$value, FUN = "min", by = list(rad$Var1))$x

par(las = 1)
plot(avgrad ~ sort(unique(dPC$L1)), pch =16, 
     xlab = "Maximum Dispersal Distance thresholds Log(m)", cex = 1,
     ylab = "dPC Standard Score (z-score)", log = "x", type = "b", lty = 2, 
     cex.lab = 1.2, cex.axis = 0.7, ylim = c(-0.4,3))
points(log(maxrad) ~ sort(unique(dPC$L1)), type = "b", col = "darkgreen",lwd = 2)
points(minrad ~ sort(unique(dPC$L1)), type = "b", col = "blue",lwd = 2)
rect(13000, -0.4, 100, 3, col = colr[1], border = NA)
rect(13000,  -0.4, 52000,-0.03, col =colr[2] , border = NA)
rect(52000, -0.4 ,240000,-0.14, col = colr[3], border = NA)
abline ( h = 0)
legend("left",c("Home Range", "Dispersal", "Migration"), fill = bars, title = "Type of movement")
legend("right",c("Log(max)", "Median", "Min"), lty = 1, col = c("darkgreen", "black", "blue"), title = "dPC z-scores")

dev.off()
```



```{r}
plot(sapply(unique(L1), function(x) median(value[L1 == x])) ~ unique(L1) , dPC, 
      cex = 1.2,  col = unique(PCnum$col), 
     pch = 15, log = "x",
     cex.lab = 0.7, cex.axis = 0.7, 
     xaxt = "n", ylab = "median dPC")
points(sapply(unique(L1), function(x) median(value[L1 == x])) ~ unique(L1) , dPC, 
       cex = 1.2, pch = 0,
       cex.lab = 0.7, cex.axis = 0.7)
```





```{r, echo = FALSE}

PCIntra <- lapply(metric ,function (x) sapply(x[,4], function(x) x)) # PCintra values
names(PCIntra) <- as.numeric(sapply(names(PCIntra), function(x)gsub( "m.csv", "",x))) # extract the "m.csv" pattern from the names 
PCIntra <- lapply(PCIntra, function(x) sort(x, decreasing = T)) # Sort PC (useful to apply colors in maps later)
PCIntra <- reshape2::melt(PCIntra) # flatten list
PCIntra$L1 <- as.numeric(PCIntra$L1) # as numeric threshold distances


PCflux <- lapply(metric ,function (x) sapply(x[,5], function(x) x)) # PCflux values
names(PCflux) <- as.numeric(sapply(names(PCflux), function(x)gsub( "m.csv", "",x))) # extract the "m.csv" pattern from the names 
PCflux <- lapply(PCflux, function(x) sort(x, decreasing = T)) # Sort PC (useful to apply colors in maps later)
PCflux <- reshape2::melt(PCflux) # flatten list
PCflux$L1 <- as.numeric(PCflux$L1) # as numeric threshold distances


PCcon <- lapply(metric ,function (x) sapply(x[,6], function(x) x)) # PCconnc values
names(PCcon) <- as.numeric(sapply(names(PCcon), function(x)gsub( "m.csv", "",x))) # extract the "m.csv" pattern from the names 
PCcon <- lapply(PCcon, function(x) sort(x, decreasing = T)) # Sort PC (useful to apply colors in maps later)
PCcon <- reshape2::melt(PCcon) # flatten list
PCcon$L1 <- as.numeric(PCcon$L1) # as numeric threshold distances
```

```{r, echo = FALSE}
library(sp)

distances <- as.numeric(gsub( "m.csv", "",names(metric)))
order <- order(distances)
names(order) <- distances[order]

```


```{r, echo  =FALSE, warning=FALSE, fig.align='center', fig.cap = "Left pane: Overall contribution (in percentage) of the different PC fractions (PCIntra, PCFlux, PCconnector) to the overall habitat availability measured by PCnum in function of the maximum dispersal distances tested on the models. Righ pane: Correlation of patch dPC fractions with the patch attribute (patch relative area). PCconnect values for all patches at 100m dispersal distances = 0, therefore no correlation was possible to calculate. Note that x axis have been logarithmized. "}
png(filename = "www/ReportFilesPlots/Hyphothesis2.4.png",width = 1000, height = 1500, res = 200  )
par(mfrow = c(2,1), las = 1, oma = c(0.1,0.1,0.1,0.1),mar = c(4,6,6,3))
colpal = rev(c('#67001f','#b2182b','#d6604d','#f4a582','#fddbc7','#d1e5f0','#92c5de','#4393c3','#2166ac','#053061'))

dthres <- as.numeric(gsub( "m.csv", "",names(metric)))
plot(sum(metric[[1]]$dPCintra)/sum(metric[[1]]$dPC)~dthres[order[1]],
     ylim = c(0,1),col = "black",pch = 16,cex = 1.3,
     xlim =c(min(dthres),max(dthres)),log = "x",
     ylab = "Metric contribution to connectivity %",
     xlab = "Maximum Dispersal Distance thresholds Log(m)")
for(i in order[-1]){
points(sum(metric[[i]]$dPCintra)/sum(metric[[i]]$dPC)~dthres[order[i]],
       col ="black", pch = 16, cex =1.3)
  points(sum(metric[[i]]$dPCintra)/sum(metric[[i]]$dPC)~dthres[order[i]],
       col ="gray30", pch = 1, cex =1.3)
  }
for(i in order){
  points(sum(metric[[i]]$dPCflux)/sum(metric[[i]]$dPC)~dthres[order[i]],
         pch = 17,  col = "black", cex =1.3)
  points(sum(metric[[i]]$dPCflux)/sum(metric[[i]]$dPC)~dthres[order[i]],
         pch = 2,  col = "gray30", cex =1.3)
  }
for(i in order){
  points(sum(metric[[i]]$dPCconnector)/sum(metric[[i]]$dPC)~dthres[order[i]], 
         pch = 15,col = "black", cex =1.3)
   points(sum(metric[[i]]$dPCconnector)/sum(metric[[i]]$dPC)~dthres[order[i]], 
         pch = 0,col = "gray30", cex =1.3)
  }  

rect(13000, -0.4, 100, 1, col = colr[1], border = NA)
rect(13000,  -0.4, 52000,1, col =colr[2] , border = NA)
rect(52000, -0.4 ,240000,1, col = colr[3], border = NA)
legend("topright", legend = c("PCintra", "PCflux", "PCconnector"), pch = c(1,2,0), horiz = F, cex =0.6)




metPCIntra <- editMetric(metric, 4) #dPCIntra in column 4
metPCflux <- editMetric(metric, 5) # PCflux in column 5
metPCConnect <- editMetric(metric, 6) # PCconnect in column 6
metCom <- editMetric(metric, 7) # Number of components in column 7 

colpal = c('#67001f','#b2182b','#d6604d','#f4a582','#fddbc7','#d1e5f0','#92c5de','#4393c3','#2166ac','#053061')

names(colpal)=  sort(metPCIntra$distances)

corr <- c()
for(i in 1:length(metric)){ 
  corr$PCflux[i] <- cor(metric[[i]]$dPCflux,metric[[i]]$dA)
  corr$PCconnector[i] <- cor(metric[[i]]$dPCconnector,metric[[i]]$dA)
  corr$PCIntra[i] <- cor(metric[[i]]$dPCintra,metric[[i]]$dA)
  corr$dist[i]<-metCom$distances[i]
  }

pal <- rev(colpal)[match(corr$dist,as.numeric(names(colpal)))]

plot(PCflux~dist, data = corr,
  col = "black", pch = 17, cex = 1.2,
  log = "x", ylim = c(0,1),
  xlab = "Maximum Dispersal Distance thresholds Log(m)", 
  ylab = "Pearson correlation with dA")
  points(PCflux~dist, data = corr, col = "gray30", pch = 2, cex = 1.3)
points(PCconnector~dist, data = corr,col ="black", pch = 15, cex = 1.2)
points(PCconnector~dist, data = corr, col = "gray30", pch = 0, cex6= 1.3)
points(PCIntra~dist, data = corr, col = "black", pch = 15, cex = 1.2)
points(PCIntra~dist, data = corr, col = "gray30", pch = 1, cex = 1.3)
rect(13000, -0.4, 100, 1, col = colr[1], border = NA)
rect(13000,  -0.4, 52000,1, col =colr[2] , border = NA)
rect(52000, -0.4 ,240000,1, col = colr[3], border = NA)
legend("bottomright", pch = c(2,0,1), 
       legend = c(names(corr[1:3])) , cex = 0.6)

dev.off()
 
```
. 



```{r, echo  =FALSE, fig.align='center', fig.cap="Variation on the dPC metric with latitude for all patches for different maximum distance dispersal scenarios. Latitude of the centroid patches are represented in the y axis and ECA values along the x axis. Horizontal dotted line represent the latitude of the Costa Rica - Nicaragua border at the Atlantic coast of the ROI, Vertical dotted line represent the maximum ECA value reached for a single patch for all distance thresholds calculated (i.e. maxECA at 1500m)"}
  par(mfrow = c(2,5), oma = c(2,1,2,1), mar = c(2,2,1,0), las = 1)
ee <- names(order)[match(names(order), names(order)[order])] # rearrange to match legend
rad
for(i in order) {
  plot((metric[[i]]$dPC),coordinates(dist)[,2], type= "l",xlim = c(0,40),yaxt = "n", ylab = "Latitude",
                              pch = 16, cex = 0.6, col = "gray70")
  points((metric[[i]]$dPC),coordinates(dist)[,2], 
                              pch = 16, cex = 0.6, col = rev(colpal)[match(distances[i],names(colpal))])
  abline(h = (10.94*1661139)/15, lwd = 1, lty =2)
  abline(v = max(metric$`1500m.csv`$dPC ), lty = 4)
  
  axis(2, c(8,9,10,11,12,13, 15), at = seq(min(coordinates(dist)[,2]),max(coordinates(dist)[,2]), length.out = 7))
 legend("bottomright", legend = ee[i], bty = "n", text.font = 2)
  }
title(" dPC by latitude", outer = T)

dPC


```


```{r}
png(filename = "www/ReportFilesPlots/Hyphothesis2.3.png",width = 1900, height = 3000, res = 250 )
par(las = 1, oma = c(3,3,3,3))
plot(scale(metric[[1]]$dPC),coordinates(dist)[,2], xlim = c(-3,20),type = "l",
       ylab = "Latitude",yaxt = "n",
       xlab = "dPC z-scores", col =bars[1], cex.lab = 1.5)
for(i in seq(2,7,1)) {
points(scale(metric[[i]]$dPC),coordinates(dist)[,2], type= "l",
                              pch = 16, cex = 0.6, col = bars[1])}
for(i in seq(8,9,1)) {
points(scale(metric[[i]]$dPC),coordinates(dist)[,2], type= "l",
                              pch = 16, cex = 0.6, col = bars[2])}

for(i in 10) {
points(scale(metric[[i]]$dPC),coordinates(dist)[,2], type= "l",
                              pch = 16, cex = 0.6, col = "blue")}


axis(2, c(8,9,10,11,12,13, 15), at = seq(min(coordinates(dist)[,2]),max(coordinates(dist)[,2]), length.out = 7))
abline(h = (10.94*1661139)/15, lwd = 1, lty =2)
legend("right",c("Home Range", "Dispersal", "Migration"), fill = bars, title = "Type of movement")
dev.off()
```






=== 
Appendix
===












<!-- ```{r, echo = FALSE, fig.align='center', warning=F, fig.cap="Cumulative sum of the dPCnum metrics for all patches at the different dispersal distance thresholds. Y axis in on a logarithmic scale to differentiate better dPCnum patterns between thresholds", warnings=F} -->
<!-- order <- match(as.numeric(gsub( "m.csv", "",names(metric))),as.numeric(names(colpal))) -->
<!-- col = colpal[order] # Match palette order with metric object order -->

<!-- par(las = 1, mfrow = c(1,3)) -->

<!-- plot(cumsum(sort(metric[[1]]$dPCintra)), -->
<!--      ylim = c(1e-07,100), xlim = c(-100, 700),  -->
<!--      log ="y", type ="l", col = col[1],  -->
<!--      lwd = 3, xlab = "Patches Rank",   -->
<!--      ylab = "∑ dPCIntra") -->
<!--   for(i in 2:10){ points(cumsum(sort(metric[[i]]$dPCintra)),  -->
<!--                          col =col[i], type = "l", lwd = 3)} # Add rest of metric to plot -->
<!-- legend("topleft", names(colpal), col = colpal, pch = 16, cex = 0.7) -->


<!-- plot(cumsum(sort(metric[[1]]$dPCflux)), ylim = c(1e-07,100),  -->
<!--      xlim = c(-100, 700), log ="y", type ="l",  -->
<!--      col = col[1], lwd = 3, xlab = "Patches Rank", -->
<!--      ylab = "∑ dPCFlux") -->
<!--   for(i in 2:10){ points(cumsum(sort(metric[[i]]$dPCflux)),  -->
<!--                          col =col[i], type = "l", lwd = 3)} # Add rest of metric to plot -->
<!-- legend("topleft", names(colpal), col = colpal, pch = 16, cex = 0.7) -->

<!-- plot(cumsum(sort(metric[[1]]$dPCconnector)), -->
<!--      ylim = c(1e-07,100),xlim = c(-100, 700),  -->
<!--      log ="y", type ="l", col = col[1],  -->
<!--      lwd = 3, xlab ="Patches Rank",  -->
<!--      ylab = "∑ dPCconnector") -->
<!-- for(i in 2:10){ points(cumsum(sort(metric[[i]]$dPCconnector)), col =col[i], type = "l", lwd = 3)} # Add rest of metric to plot -->
<!-- legend("topleft", names(colpal), col = colpal, pch = 16, cex = 0.7) -->
<!-- ``` -->







```{r, eval = FALSE,  echo = FALSE, out.width='400px', warning=F, results=F, fig.align="center", fig.cap="Summary plot with the results for the connectivity models at different dispersal thresholds in Raphia taedigera patches. The right pane shows different statistical metrics sumarizing the dPC (Probability of connectivity) metric values for all patches. At the topright panel the relative partitioning of the network of patches due to clustering is expresed as the ratio: number of components/number of patches. Values ranges from 0-1 being 1: n components = n patches, 0:  n components = 1 (i.e. all patches in the network are connected). Bottomright panel shows the maximum dPC value of the distribution of dPC values at each dispersal thresholds. Shadowed rectangles highlight the portion of the simulations for which the dispersal thresholds correspond to actual literature observations of dispersal distances in Tapirs "}
## Plot PC METRIC summary values (with the excemption of max values)

split.screen(rbind(c(0.15,0.6,0.05, 0.98), c(0.6, 0.98, 0.05, 0.6), c(0.6, 0.98, 0.6, 0.98), c(0.0, 0.13, 0.1, 0.98), c(0.01,0.9,0.01,0.2)))
cex = 1
colr = scales::alpha("#a7c78e",alpha = 0.3)
screen(1)
par(las = 2, mar = c(4,4,0,1), oma = c(1,1,1,1))
plot(Min~distances, data = metPC, 
    ylim = c(0,0.4), log = "x", pch = 8, 
    col =colpal, cex = cex, 
    ylab = "dPC metrics ", xlab = "Dispersal thresholds (meters)", cex.lab = 0.7, cex.axis = 0.7)
points(firstQ~distances, metPC,
        pch = 10,col =colpal, cex = cex)
points(Median~distances, metPC,
        pch = 15, col =colpal , cex = cex)
points(Mean~distances, metPC,
        pch = 16, col =colpal, cex = cex)
points(thirdQ~distances, metPC,
        pch = 6, col =colpal, cex = cex)
rect(1500, 0,52000,0.35, col = colr, border = NA)
# Add trend lines

#for(i in 1:length(tline)){abline(tline[[i]], lty =2, col=colpal[i]) }
screen(2)
par(las=2, mar = c(4,4,0,0))
# Plot max values
plot(Max~distances, metPC, 
      log ="x",  pch = 16,
     ylab = "Max dPC", xlab = "Dispersal thresholds (meters)", 
     col =colpal, cex = cex, ylim = c(0,35), cex.lab = 0.7, cex.axis = 0.7)
rect(1500, 15,52000,35, col = colr, border = NA)
screen(3)
par(las=2, mar = c(0,4,0,0))
# Plot number of components 
plot(Max/730~distances, metCom, 
     log ="x",  pch = 16,
     ylab = "# components / # patches ", xlab = "Dispersal thresholds (meters)", 
     col =colpal, cex = cex, ylim = c(0,1) ,xaxt = "n",cex.lab = 0.7, cex.axis = 0.7)
rect(1500, 0,52000,0.2, col = colr, border = NA)
screen(4)
par(mar = c(0,0,0,0), oma = c(0,1,2,0))
plot.new()
legend("center", legend = c("Min", "1st Q", "Median", "Mean", "3rd Q"),
       pch = c(8,10,15,16,6), title = "Metrics")

```

### Number of components 


```{r, echo = FALSE, fig.align="center", fig.cap="Ratio between the number of components and total number of patches calculated for different dispersal threshold values. Number of patches is fixed overall calculations, therefore, variations on this ratio represents the rate of clustering and connectivity among patches when different dispersal distance thresholds considered. Ratios equals to one (1) represent a complete unconnected network of patches (i.e. No effective connection (dispersal) among any pair of patches) Highlighted within the green rectangle are the values for which dispersal thresholds correspond to the tapir dispersal distances reported in literature."}
par(las = 1)
plot(Max/730~distances, metCom, 
     log ="x",  pch = 16,
     ylab = "No. components / No. patches ", xlab = "Dispersal thresholds log(meters)", 
     col =colpal, cex = 1.5, ylim = c(0,1),cex.lab = 1)
rect(1500, 0,52000,0.2, col = colr, border = NA)
legend("topright", legend = sort(unique(dPC$L1)),
       pch =16, col = colpal[order(as.numeric(names(colpal)))], title = "Dispersal thresholds", bty = "n")
```