Why is the effect of trait variation on population persistence larger in cyanobacteria than in sulphur bacteria?

[romanalimb]

I am here comparing the response to oxygen diffusivity of CB when their environment deteriorates (i.e. when oxygen diffusivity decreases) to the response of sulphur bacteria when their environment deteriorates (i.e. when oxygen diffusivity increases):

When the environment deteriorates, populations with trait variation crash later than populations without trait variation. This effect of diversity is of larger magnitude for cyanobacteria than for sulphur bacteria (Fig. 7a vs. Fig. 9 b+c). Why?

Some differences between the functional groups that could contribute to these different patterns: • Differences in the nonlinearity of the inhibiting compounds (also see issue # 9). • Trait differences (the highest maximum growth rate in CB is lower than the highest maximum growth rates in SB and PB) • SB depend on the production of SO by PB, and PB depend on the production of SR by SB. No such interaction exists for the CB.

[opetchey]

Related to point 2: We know that decreasing the amount of trait variation in CB does reduce the amount of delay in crash during deterioration (figure 11 pink). We don't know what would happen to the SB/PB deterioration delay if they had even more variability than in experiment 1. This is something to investigate.

[opetchey]

I'm running the sims with additional SBPB variation...

[romanalimb]

We noted that when not logarithmizing oxygen diffusivity, the effect of trait variation on population persistence is in fact larger for the sulphur bacteria than for the cyanobacteria. But the question remains: why does the magnitude of the effect of trait variation on population persistence differ among functional groups?

[opetchey]

The new pdf for experiment 1 has a new section, extra variation in SBPB. It shows that when we put more variation/diversity into these groups, there can be just as much effect of diversity. Note that the process for adding trait variation is multiplicative, so if the gmax is lower in one group, then the absolute variation will be lower, when multiplicatively the same. And I think it must be the absolute trait value that is important.

[opetchey]

I suggest to close this issue, or to clearly specify next steps.

[opetchey]

I do think it is worth comparing the effect of diversity when the absolute amount of diversity is the same (or is as similar as possible).

[opetchey]

Well, I did this, and didn't learn a great deal more than we already knew. It turned out to not be possible (or at least I couldn't) to make the absolute variation in realised growth rate the same for CB, SB, PB. Anyway, I'm re-running the big sim with greater variation in SBPB, and a few other tweaks, so we have those results too, which will more clearly show that relative variation is what is important, and that the SBPB diversity effect can be as large as the CB diversity effect.

[opetchey]

@romanalimb sims rerun with more variation in CB and also SBPB. Results are in the section 2x CB, 6xSBPB diversity of this file: https://github.com/UZH-PEG/diversity_envresp1/blob/main/experiments/experiment%201/experiment-1.pdf

The results are pretty much what I expected... with more SBPB variation we can get a larger effect than with CB variation. And variation in the two groups have interesting effects, with them both tending to have a larger effect extending the resilience of the groups to deterioration in their environment.

SBPB does cause earlier shift when their environment ameliorates, but as we previously discussed, this is a bit of a puzzle given that we see no strain replacement with SBPB as their environment ameliorates. This puzzle needs to be resolved!

PS. the two new datasets are in the Teams channel files.

[romanalimb]

@opetchey I opened a separate issue for the SBPB amelioration puzzle.