Beta-diversity
Abstract
These methods are used to measure the beta-diversity of two networks, by partitioning variation into the dissimilarity of species (βS), interactions (βOS), and whole networks (βWN).
Components of networks beta diversity
#
SpeciesInteractionNetworks.BetaDivComponent — Type.
BetaDivComponent
The betadiversity methods all use a subtype of BetaDivComponent as their first argument, to determine which component should be measured.
All of the partitions follow the Koleff et al. (2003) approach, where the beta diversity is measured on the cardinality of sets. Specifically, all betadiversity functions will return a named tuple with three fields, called shared, left, and right. These represent, respectively, the number of items common to both networks, the number of items unique to the first argument, and the number of items unique to the second argument.
References
Koleff, Gaston and Lennon (2003)
#
SpeciesInteractionNetworks.βS — Type.
βS
The species dissimilarity between two networks is, straightforwardly, the usual β diversity applied to the list of nodes. In the case of networks that are not uniparite, we can further calculate this dissimilarity for the different dimensions of the network.
References
Koleff, Gaston and Lennon (2003)
#
SpeciesInteractionNetworks.βOS — Type.
βOS
The overlapping-species interaction dissimilarity measures the dissimilarity in the interactions between species that are shared between two networks. In other words, interactions are only compared if they involve two interactions that are established between species present in both networks.
References
Poisot, Canard, Mouillot, Mouquet and Gravel (2012)
Canard, Mouquet, Mouillot, Stanko, Miklisova and Gravel (2014)
#
SpeciesInteractionNetworks.βWN — Type.
βWN
The whole-network interaction dissimilarity measures the dissimilarity between interactions in all species of either network. It is trivially maximized when the networks have no species in common.
References
Poisot, Canard, Mouillot, Mouquet and Gravel (2012)
Canard, Mouquet, Mouillot, Stanko, Miklisova and Gravel (2014)
What about the species turnover component?
In Poisot et al. (2012), we introduced the idea that the impact of species turnover can often be expressed as the difference between the whole-network (βWN) and overlapping-species (βOS) dissimilarities. This is only true for some measures. Following the arguments laid out in Poisot (2022), we have not added this as a built-in function. If there is a need to measure the impact of turnover, it is recommended to express it as (wn-os)/wn.
Beta diversity measures
The betadiversity function will always return a named tuple with three entries, named
#
SpeciesInteractionNetworks.betadiversity — Function.
betadiversity(::Type{βS},U::T,V::T,dims::Integer = 0,) where {T <: SpeciesInteractionNetwork{<:Partiteness, <:Binary}}
Species-level β diversity between networks. By default, this return the species dissimilarity for the entire network. An optional last argument dims can be used, to specify top (1) and bottom (2) levels.
betadiversity(::Type{βWN},U::T,V::T) where {T <: SpeciesInteractionNetwork{<:Partiteness, <:Binary}}
Network-level β diversity between networks. By default, this return the species dissimilarity for the entire network. An optional last argument dims can be used, to specify top (1) and bottom (2) levels.