Background sampling.Rmd

---
title: "Background sampling for ENMTools Models"
author: "Dan Warren"
date: "8/12/2020"
output: html_document
---


```{r}
library(raster)
library(ENMTools)
```


# Implicit choice of background (points > range > env)

Creating a species object with no background
```{r warning = FALSE, message = FALSE}
monticola <- iberolacerta.clade$species$monticola
monticola$range <- NA
interactive.plot.enmtools.species(monticola)
```


If no range raster or background points exist, ENMTools just samples background points from the full env extent.
```{r warning = FALSE, message = FALSE}
monticola.gam <- enmtools.gam(monticola, euro.worldclim)
interactive.plot.enmtools.model(monticola.gam)
```


If we add a range raster it defaults to sampling background from that instead.

```{r warning = FALSE, message = FALSE}
monticola$range <- background.raster.buffer(monticola$presence.points, 100000, euro.worldclim)
monticola.gam <- enmtools.gam(monticola, euro.worldclim)
interactive.plot.enmtools.model(monticola.gam)
```


However if we add background points to the species object, it will default to using those first.  Here I'm sampling some background points from a raster of population density across this area of Europe.

```{r warning = FALSE, message = FALSE}
europop <- raster("~/GitHub/ENMTools Tutorials/europop.asc")

pop.points <- rasterToPoints(europop, xy = TRUE)
colnames(pop.points) <- c("lon", "lat", "pop")
pop.points <- as.data.frame(pop.points)

# Just dividing the populations by the maximum so they range 0-1
pop.points$pop <- pop.points$pop/max(pop.points$pop)

# Sampling in proportion to population density
bias.points <- pop.points[sample(1:nrow(pop.points), 1000, prob = pop.points$pop),]
plot(europop)
points(bias.points, pch = 16)

monticola$background.points <- bias.points
monticola <- check.species(monticola)
monticola.gam <- enmtools.gam(monticola, euro.worldclim)
interactive.plot.enmtools.model(monticola.gam)
```


# Explicit choice of background using bg.source

Okay now let's look at the species object.
```{r warning = FALSE, message = FALSE}
monticola
```



It has presence points, background points, and a range raster.  We've been using ENMTools implicit ranking of those things til now (points > range raster > env raster), but we can call them manually as well!  We just have to supply the appropriate argument to **bg.source**.

```{r warning = FALSE, message = FALSE}
monticola.gam <- enmtools.gam(monticola, euro.worldclim, bg.source = "range")
interactive.plot.enmtools.model(monticola.gam)

monticola.gam <- enmtools.gam(monticola, euro.worldclim, bg.source = "env")
interactive.plot.enmtools.model(monticola.gam)

monticola.gam <- enmtools.gam(monticola, euro.worldclim, bg.source = "points")
interactive.plot.enmtools.model(monticola.gam)
```