SpadeR:::ChaoSpecies() error in boxplot function

[mobiodiv]

SpadeR:::ChaoSpecies() error in boxplot function

[dmcglinn]

I've also been getting some additional errors from the SpadeR:::ChaoSpecies function. This function is used to estimate the asymptotic size of the species pool. I'm not sure why it is giving us so much trouble because we do something similar in the null model portion of the analysis in which we generate the species pool's SAD. My feeling for the time being is that you should go in and just comment out any of the computations of extrapolated S. This was not a key interest of the working groups as we all agree that S never truely saturates - the concept is only useful in our opinion as a means to an ends for generating the SAD a for a pooled set of plots. @FelixMay did someone in the group encourage you to add those computations?

[rueuntal]

Is this only a problem in boxplot, not in the main function?

[dmcglinn]

yea just the boxplots

[dmcglinn]

@FelixMay wrote a function called mob_stats which computes a ton of useful statistics which are then fed into the boxplot function. Within that mob_stats function is where the trouble is.

[rueuntal]

OK I'll leave it to @FelixMay then :)

Just by reading the code I suspect the problem occurs because here Chao estimator is applied to plot-level data, while in our null test Chao's method is applied to the overall SAD lumped across all plots and all groups, which is less prone to have an extreme distribution as @valentinitnelav reported.

[dmcglinn]

Yea I think you're correct. I tried to rescue it from errors but requiring plot N be greater than 4 but likely there are other criteria that will need to be met as well. It would be nice if there was an easy way to try to compute the ChaoS for a plot and if there is no error then assign it otherwise give it an NA. Now if it hits an error the function just stops.

[rueuntal]

Hmmm do we want to remove the bit or try to fix it? If we want to fix it I can dig into Chao's paper to what they did exactly.

[dmcglinn]

I don't think that's necessary. I was thinking that this could possibly be accomplished programmatically using a function like try.

[rueuntal]

Sounds good.

[FelixMay]

Should be fixed with #132