bipartite_stats() R function from [econullnetr]

Test for significant differences in a range of network metrics between the observed and null bipartite networks

Acts as a wrapper for the bipartite package's networklevel, grouplevel and specieslevel functions, allowing a wide range of measures to be calculated (Dormann et al., 2008, 2009; Dorman 2011). These are calculated both for the observed network and across the iterations of the null model, allowing for simple tests of whether the observed values differ from those expected by chance.


bipartite_stats(
  nullnet,
  signif.level = 0.95,
  index.type,
  indices,
  prog.count = TRUE,
  ...
)

Arguments

nullnet: An object of class 'nullnet' from generate_null_net
signif.level: An optional value specifying the threshold used for testing for 'significant' deviations from the null model. Defaults to 0.95
index.type: String specifying which function to call from the bipartite package: specieslevel, grouplevel or networklevel, according to the level at which the statistics are required. For specieslevel, statistics are calculated for both 'higher' and 'lower' levels i.e. level = "both" and cannot be changed: adding a level argument will produce an error: statistics from higher and lower levels are easily separated in the output from bipartite_stats.
indices: Vector listing the bipartite network statistics to calculate. All indices are currently supported, with the exception of the dependence matrix: if ALL is specified, it will default to ALLBUTD.
prog.count: A logical value specifying whether the progress count should be shown. Defaults to TRUE.
...: Other arguments that may be supplied to bipartite's specieslevel, grouplevel or networklevel functions.

Returns

Returns one or more data frames according to the level at which the statistics are calculated (specieslevel, grouplevel or networklevel). If index.type = "networklevel" or index.type = "grouplevel" a single data frame is returned, listing the chosen network statistics and with the following column headings:

Observed: Value of the statistic for the observed network
Null: Mean value of the statistic across the iterations of the null model
Lower.CL: Lower confidence limit for the metric
Upper.CL: Upper confidence limit for the network metric
Test: Whether the value of the statistic with the observed network is significantly higher than expected under the null model, lower or consistent with the null model (ns )
SES: The standardised effect size for the difference between the observed network and the null model (see Gotelli & McCabe 2002 for details)

If index.type = "specieslevel", a list comprising two data frames for each statistic, representing the higher and lower levels in the network. Each data frame has the same format as for networklevel except that the rows are individual nodes (species) in the network. See examples for how to call the individual data frames.

Details

Allows most of the network metrics in the bipartite package to be calculated for an observed bipartite network and compared to the distribution of those network metrics across the iterations of the null model. This indicates whether the observed network differs from the structure of the network that could be expected if consumers simply used resources in proportion to their relative abundance.

The user sets the significance level (default = 95% confidence limits), and the metrics selected are classified into those that are higher, lower or consistent with the null model at that significance level. Significance is determined by comparing the observed value of the statistic to the 1-alpha/2 percentiles from the frequency distribution, with 'significant' values falling outside the confidence interval (Manly 2006).

Examples


set.seed(1234)
sil.null <- generate_null_net(Silene[, 2:7], Silene.plants[, 2:6], sims = 10,
                              c.samples = Silene[, 1],
                              r.samples = Silene.plants[, 1])
# Network-level analysis
net.stats <- bipartite_stats(sil.null, index.type = "networklevel",
                             indices = c("linkage density",
                             "weighted connectance", "weighted nestedness",
                             "interaction evenness"), intereven = "sum")
net.stats

# Group-level analysis
grp.stats <- bipartite_stats(sil.null, index.type = "grouplevel",
                             indices = c("generality",
                             "vulnerability", "partner diversity"),
                             logbase = 2)
grp.stats

# Species-level statistics
spp.stats <- bipartite_stats(sil.null, index.type = "specieslevel",
                             indices = c("degree", "normalised degree",
                             "partner diversity"), logbase = exp(1))

spp.stats # Show all data frames of results
spp.stats$normalised.degree # Select one statistic
spp.stats$normalised.degree$lower # Select one statistic at one level

References

Dormann, C.F., Gruber B. & Frund, J. (2008). Introducing the bipartite package: analysing ecological networks. R news, 8 , 8--11.

Dormann, C.F., Frund, J., Bluthgen, N. & Gruber, B. (2009). Indices, graphs and null models: analyzing bipartite ecological networks. Open Ecology Journal 2 , 7--24.

Dormann, C.F. (2011) How to be a specialist? Quantifying specialisation in pollination networks. Network Biology, 1 , 1-20.

Gotelli, N.J. & McCabe, D.J. (2002) Species co-occurrence: a meta-analysis of J.M. Diamond's assembly rules model. Ecology, 83 , 2091--2096.

Manly, B.F.J. (2006) Randomization, Bootstrap and Monte Carlo Methods in Biology (3rd edn). Chapman & Hall, Boca Raton.

Vaughan, I.P., Gotelli, N.J., Memmott, J., Pearson, C.E., Woodward, G. & Symondson, W.O.C. (2018) econullnetr: an R package using null models to analyse the structure of ecological networks and identify resource selection. Methods in Ecology and Evolution, 9 , 728--733.

bipartite_stats function