CODEWITHSUNDEEP
rvegan
cj456
cj456

Reputation: 11

envfit output: arrow direction interpretation

I'm using envfit() in R and I don't understand the output. While some people don't value arrow directionality, others say that arrow directionality is important.

I know that the longer the arrow, the stronger the influence. But does that mean that the samples in that direction are affected by that factor?

Upvotes: 1

Views: 493

Answers (1)

Jari Oksanen
Jari Oksanen

Reputation: 3682

The following should explain the direction of the arrow:

library(vegan)
data(mite, mite.env)
m <- rda(decostand(mite, "hellinger"))
plot(m, display="site")
## show arrows for two continuous variables
plot(envfit(m ~ WatrCont + SubsDens, mite.env))
## Add linear trend surface for WatrCont
ordisurf(m ~ WatrCont, data=mite.env, knots=1, add=TRUE)

Linear trend surface shows the fitted values of the variable WatrCont and the arrow shows the gradient of the slope of that trend surface. The arrow is perpendicular to the isolines of the trend surface, and the arrow shows the direction of the most rapid change (slope) of the surface (and the most rapid decrease would be an opposite arrow). So yes, the arrow indicates that the fitted values of the variables are higher to the direction of the arrow. Please note that you must project points to the arrow, that is, to follow the isolines of the trend surface, and it is not the point of the arrow, but the direction that counts.

Arrows are a nice and compact way of displaying the environmental variables when you (can) assume a linear trend for those variables.enter image description here

Upvotes: 3

Related Questions