kaitlyngaynor
commented
9 months ago Hmm maybe not worth adding, as we've got huge margins of error. I think the table says enough, and you can speak to the large standard error in the discussion and indicate that the evidence for interaction is therefore weak, even though it does appear in the best model
https://github.com/kaitlyngaynor/gorongosa-mesocarnivores/blob/3064bb17fa87e0fe8ce145d70d58e6b97e432d47/scripts/species-interaction-model/02-multi-species-occupancy-analysis-GNP-2023.Rmd#L629
Hi! You said this in your email, re: the civet:mm new result "Can you plot their co-occurrence somehow, based on the model predictions? I’m not familiar enough with this modeling approach to know how to generate a figure like this but presume it’s possible. The standard error is HUGE tho so I imagine you’ll have some big confidence interval bands."
from this page, I followed the "plotting conditional occupancy" section and came up with the following plots:
civet occupancy given mm present/absent:
mm occupancy given civet present/absent:
I haven't dug further to see if there are other options for plotting because I think this is what you were talking about. but as you said, there are very large confidence intervals.
what do you think about including one of these? I made both because there's not as clear a dominant species as in the example from the site, though I'm pretty sure civets would be dominant in this pair (so the second figure would make more sense to include)