Network β-diversity

Measures of β-diversity work by first calculating the unique/shared items (using the βs, βos, and βwn functions), then passing on these arguments to one of the KGLXX functions to return a (dis)similarity score. The KGL functions are named for Koleff, Gaston, and Lennon -- the number of each function matches the number in Table 1.

β-diversity components

The package implements functions for the βs, βos, and βwn components of network dissimilarity. In the original publication, we also described βst, which was the proprotion of dissimilarity due to species turnover, and defined as βst = βwn - βos for measures of dissimilarity bounded between 0 and 1. After discussing with colleagues and considering our own use-cases, it appears that the interpretation of βst is not always straightforward, and so we have decided to exclude it form the available functions.

```@docs βs βos βwn

One measure which is possibly missing is a function to build the metaweb, *i.e.*
aggregate all species and interactions in a collection of networks. This can be
done using `union`, *e.g.*

~~~julia
N = nz_stream_foodweb() # Or any collection of networks
metaweb = reduce(union, N)
~~~

## β-diversity measures

```@docs
KGL01

Basic operations on networks

Internally, the functions for β-diversity rely on the usual operations on sets. The act of combining two networks, for example, is a union operation.

@docs setdiff union intersect