Introduction to Species Distribution Models

Bob O'Hara
Sept 11 2015

Where is this species?

TTWP data

Why is it there?

TTWP data

What we want

A prediction of its distribution

Estimation of the factors affecting the distribution

But why is the species there?

The environment!

$Some Environment$

So, we need to make the link

$Some Environment$

Observations -> Environment -> Distribution

To put it another way

We observe data in geographic space,

want to model this in environmental space,

and then predict back into geographic space

Two sides of the same coin

Species Distributions - in geographic space

Niche - in environmental space

The BAM model

The distribution of the species is determined by:

Biotic
Abiotic
Movements

Most SDMs only deal with the Abiotic

Fundamental and Realised Niches

Fundamental = Abiotic

Realised = Biotic (and perhaps Movement)

What's the role of space?

Most SDMs ignore it in the model: it just (!) provides a convenient way of representing the distribution

The Mathematical Model

Is a species there or not?

Model as probability, as not certain

So,

\[ Pr(I=1) = f(env) \]

Complexities

We don't have perfect data, thus we need $ Pr(X=1|I) $

This will depend on the type of data

Different data means different models

The Environmental Niche

Can model $ f(env) $ in many ways

Linear
Spline
Neural Nets
Eccentric Landing Platforms for Pixies
etc etc

What we Usually Model

How the observed distribution is predicted by the environment.

Environment -> Fundamental Niche

BUT

We observe the realised niche

And Now

… we can have a simple tutorial