Can you tell more about the protocol? If you are sure that these bees correspond to different individuals (eg. recorded at the same time), why not simply use abundance at location X instead of presence? If you are not sure, then there is a risk of double counting and this can induce a bias. What is the approximate size of each location? Did all observations have been recorded at the same date?

Hi Laura Henckel

Data were collected in savanna, grazing and agricultural landscapes of Northern Tanzania. The grazing and agricultural landscapes were divided into two levels (mid/high) by using QGIS making a total of five landscapes. In each landscape eight 50x50m plot were established of which bees were collected by using pan traps (left on site for 48 hrs) and sweep netting.

Each plot was surveyed on different date therefore the bees collected corresponds to different individuals. Also, I'm don't know how to use abundance in SDM.

Hope I answered your concerns

Abraham

Abraham; Except for these two bees, it sounds like all the other pans got exactly one individual of species X.. Something is remarkable about that. Is that actually what the data say? Jim Des Lauriers

Hi James Des Lauriers

I did not got clearly your question, however the data has different species and I was interested to do SDM for one species.

Abraham

Abraham; When I analyze my own ant data about similar questions (not using SDM) the numbers if each species at a site are pooled. Then the numbers from different landscapes are entered into the analysis. Is there a restriction in your SDM algorithm about pooling data? If not, that is what I would do. Best regards, Jim Des Lauriers

Hi James Des Lauriers

There is no restriction in my SDM regarding pooling data of one species from different landscapes. Also, my SDM will include only one species of bee.

Abraham

Abraham; Well then, pool those two records of species X and run the analysis. That's a happy outcome. Jim Des Lauriers

If you want to use abundance instead of presence/absence, you can use glm function with poisson distribution and log link. See an example here: https://online.stat.psu.edu/stat504/node/169/

If you want to use abundance instead of presence absence data, you should read up on species abundance distribution models. Lots of useful information. But to answer your question: you should remove any duplicate locations, because the algorithm you train will be tricked into thinking that the variables at this site with duplicates will hold more importance than other variables, and therefore create an output that is bias towards spesific variables. Hope this makes sense.