timeseries analysis

[emchristensen]

the main question is how the community changes over time:

• do a basic ED between time steps of the community

[beecycles]

Multivariate analysis options to determine relationship between ant/plant/rodent community compositions for time series data:

• CoCA: Co-Correspondence analysis. Takes two species community datasets at a time and can be run as a symmetric CoCA to determine patterns common to both communities over time, or as a predictive CoCA where one community is the response variable and you test how well the composition of one taxa predicts the composition of another at a given time. See more here.

• [ ] 1. Organize necessary data: species matrices for ants, plants, and rodents with species as columns, year as rows, and abundances as cell values @emchristensen

• [ ] 2. Run and explore how useful these results might be @beecycles
  

• PRC, derived from RDA Tracks changes in community composition for time series data, and can incorporate plot treatment into analysis. Uses vegan package in R with following syntax: rda(response ~ treatment * time + Condition(time)). See more here. Plot below is an example of output, with effect as treatment and species names listed on right side in order of species scores (how much they are influenced by treatment I think). Need to read more on this, but just including what I've got so far. Effect could maybe be krat abundance instead of treatment to more directly look at the relative influence of krats on ant species over time.

• [ ] 1. Organize necessary data: I think it will use same matrix as above, maybe with an additional matrix corresponding to treatment or krat abundance (effect variable), not sure yet. @emchristensen

• [ ] 2. Run and explore how useful these results might be @beecycles

• Next: also look more at PCNM and asymmetric eigenvector maps (AEM) (models) as discussed here. @beecycles