dmcglinn
commented
3 years ago I guess the other thing to note here is that each grid cell gets assigned an urban cover value that is the average of the checklist values. I haven't checked if ghm values are assigned per checklist or per grid cell. If they are assigned per checklist then for this multiscale analysis that isn't ideal and we'd rather have a more representative grid cell ghm that has nothing to do with where the bird checklists were collected (I think). Let's chat more about this.
Controlled for the number of checklists per sampling grid cell - set at 50 for this first pass. Computation is relatively slow so output of analysis saved as .Rdata file.
Note in the graphics that effort is shown in two ways. In the first version of the multi-scale figure effort refers to number of checklists for the sSBR but number of individuals for the nsSBR and IBR. In the second plot all the graphics are rescaled to number of individuals. The reason for including both here is just to get a full look at the scale dependent bivariate relationships that go into the scale dependent signiature of the slopes for aggregation and SAD effects.
This is the first dataset that I've analyzed with such a strong aggregation effect. Essentially what this says is that as urban cover increases that birds are more strongly spatially clustered leading to fewer species than you would expect under random sampling. This effect is the most detectable at intermediate levels of effort but note that the aggregation effect must go to zero at large number of samples because when all samples are collected the sSBR must equal the nsSBR so agg effect (sSBR - nsSBR) has to go to zero. I'm not sure why the null model looks so screwy - It should in theory be right around zero slope. This is just for 19 permutations but it may also reflect the pecularity of the spatial sampling algo I'm using. We have a slower more intuitive one I can also try.