2.9 | Null modelling between communities
The same principles which underlie the use of phylogenetic null
modelling within a given community, can be used to infer possible
mechanisms for the variation observed between communities. For between
community comparisons, nearest taxon distances are instead calculated
between an ASV in one microbiome and its closest relative in a second
microbiome (β mean nearest taxon distance [βMNTD]) (Stegen et al.,
2012). As above, we standardized βMNTDs by the mean and standard
deviation of a null distribution created via 999 randomly assembled
community pairs (βMNTDses). Positive
βMNTDses indicate that two communities are more
phylogenetically disparate than would be expected given random sampling
of a given pool of gamma diversity. Conversely, negative
βMNTDses indicate that two communities are more
phylogenetically similar than would be expected. Assuming taxa
niche-spaces and phylogenies are correlated, then positive and negative
βMNTDses values can indicate that the differences or
similarities observed between two communities might be the result of
differential or similar selective pressures, respectively (Stegen et
al., 2012). |βMNTDses| values of
< 2 are conventionally considered to indicate that
inter-community differences might be the result of dispersal patterns or
ecological drift, as phylogenetic patterns observed between communities
do not differ greatly from those of randomly assembled community pairs.
We analyzed βMNTDses using a PERMANOVA parameterized
identically to the β-diversity analyses described above. Additionally,
we ran a mantel test to test for a correlation between spatial
separation and βMNTDses values, and a univariate
PERMANOVA to test for an effect of social band. For all nearest taxon
analyses, we used a phylogenetic tree made ultrametric (λ = 1) using the
chronos() function from the R package ‘ape’.
Finally, we also used a phylogeny-independent extension of this null
modeling framework by calculating Raup-CrickBray(RCbray) values (Chase et al., 2011; Stegen et al.,
2013). Rather than consider greater- or less- than-expected phylogenetic
similarities between communities, RCbray values indicate
whether taxa co-occur at similar abundances more or less often than
expected independent of their phylogenetic relatedness (Lowe & McPeek,
2014; Stegen et al., 2013). Among communities which do not show strong
phylogenetic deviations from null expectations, RCbrayestimates < -0.95 indicate that taxa co-occur at similar
abundances between communities more frequently than would be expected by
chance alone, one indication of homogenizing dispersal.
RCbray estimates > 0.95 indicate that taxa
co-occur between communities less often than would be expected given
random expectations, indicating dispersal limitation. Finally,
|RCbray| < 0.95 indicate
that rates of taxa co-occurrence do not differ from null expectations,
suggesting possible ecological drift. Null distributions used to make
comparisons were created via 9999 community pairs created through
randomization. Like βMNTDses, RCbrayvalues were analyzed via PERMANOVA and a partial Mantel test was used to
test for a correlation between RCbray values and
longitudinal separation, after controlling for βMNTDsesvalues.