A number of earlier posts have discussed food webs of the Permian–Triassic of the Karoo Basin in South Africa. This terrestrial ecosystem was subjected to the devastating end Permian mass extinction. The community which emerged in the aftermath of the extinction, the Lystrosaurus Assemblage Zone (LAZ), has been identified as having very unusual food web dynamics. This first figure compares the CEG dynamics of the end Permian Dicynodon Assemblage Zone (DAZ), LAZ, and the successive Cynognathus Assemblage Zone (CAZ). The implication is that there was a breakdown of perturbation dynamics during and/or right after the extinction episode. LAZ differs from the other communities (and in fact from every other community that we’ve studied so far!) in two ways:
- Levels of secondary extinction can be extremely high at low peturbation levels, implying food webs of very low resistance.
- Many species level networks or food webs (SLNs) of LAZ are nevertheless quite resistant, and resemble SLNs from the other communities. So the SLNs, or at least their dynamics, are highly variable in LAZ.
So what causes all this?
The first question we asked ourselves was, is LAZ an unusually bad community or metanetwork, or are the other Karoo communities just exceptionally good? Our approach to addressing this was to generate 1,000 random metanetworks by randomly selecting observed guild richnesses from among our observed communities to fill the richness of a random community. A random community or metanetwork could therefore have guild richnesses that never occur together in any of the observed communities, but every guild richness of a random community is observed in at least one real Karoo community. We then simulated perturbation of 100 SLNs for each random community, and collected data on the first observation above, i.e., the variability of resistance at low levels of perturbation. As we see in the second figure, LAZ really stands out, even among the random communities! Why?
Well, in order to address that, we’ve used a number of regression models to examine the dependence of that variability on proportional guild richness. Proportional guild richness, in contrast to absolute, is the fraction of a community’s total consumer richness encompassed by a particular guild. Several guilds consistently stand out: very large amphibians, very small herbivorous amniotes, very small carnivorous/insectivorous amniotes, small carnivorous/insectivorous amniotes, carnivorous insects, and herbivorous insects. Multiple regression models demonstrate that the herbivorous guilds affect resistance variability negatively, i.e., they dampen the variability, while carnivorous guilds affect it positively! Now here’s the neat part. If we examine the sub-metanetworks of DAZ, LAZ and CAZ comprising these guilds only (see figure), we can immediately see how the communities differed with respect to these crucial guilds. Guilds with a dampening effect are shown in blue, those in red have the opposite effect (producer guilds are brown). And if you think of LAZ as being somehow imbalanced or out of whack, the figures should suggest to you some ways to “restore the balance”. I’ll discuss those in the next post.