1

I have three 'variables':

  1. 6000 GPS points that represent where my focal species (a carnivore) is present. Attached to these points is one value representing the activity level of the animal at the time (it ranges between 0 and 200, thus 0 being inactive and 200 being very active).
  2. Slope (standard slope raster)
  3. NDVI (also raster)

I ran a GLMM in R Studio, and the fixed effects show that as NDVI and Slope decrease, activity decreases. This is important because it may also help with an additional analysis to assess areas of risk for my other focal species (an herbivore that also happens to be a prey species of my focal carnivore).

I am curious that given such a relationship (between activity points, NDVI, and Slope) whether it is possible to map out 'hotspots of activity' throughout the landscape while accounting for NDVI and Slope as a means to create a 'risk map' or 'probability of activity map' (same thing in this context)

I have been reading about several methods including kriging, hotspot analysis, Residual trend mapping, etc. but I am afraid that I am still unsure whether either of these methods are the right way to go (probably because I am still relatively new when it comes to GIS).

Would anyone be able to provide some advice on how to carry on with this?

  • 2
    I think you should start by turning the 6000 points to a raster surface with the activity level as the raster value. After all, your carnivores are a continuos surface and not discrete data, they don't appear and disappear like magic. – Sergio C. Jun 29 '17 at 23:56
  • 1
    Then, with the 3 raster you would be able to start doing map algebra and spatial statistics. – Sergio C. Jun 29 '17 at 23:58
  • thank you for your advice, when you mean map algebra, do you mean taking the fixed effects from the GLMM that I ran and plugging it into the raster calculator? (i.e. ndvi*.304+slope*123 etc.) or do you mean a different method. If so, I am not sure whether that would account for the actual activity points. – Redskies421 Jun 30 '17 at 9:52
  • Thank you! Yes, that's what I meant. You would like to test the similarities between the three surfaces as you were saying. – Sergio C. Jun 30 '17 at 15:24
  • Different methods to compare raster are listed here: researchgate.net/post/How_to_statistically_compare_two_maps – Sergio C. Jun 30 '17 at 15:25
2

There are a number of things you could do with this depending on your research focus. I would take Sergio's advice in the comments (and your own reading on this that you mentioned IS relevant) and start by converting the point dataset into a raster. I would do this using an interpolation. If you want to get into the Kriging that you mentioned, that is a great way to go but even something simpler like IDW can be done with spatial analyst (Kriging is an artful science that can be done well with a lot of practice). The interpolation will give you a density based estimate of "presence" of your species.

As Sergio mentioned, once you have that raster surface you have a bunch of options. You can compare the density values with the slope raster (using the raster calculator) to statistically test if slope may deter presence and likewise you could test with the NDVI to see if a high vegetation index is correlated with higher presence.

You might come up with some sort of an index that weighs both slope and NDVI in some (potentially arbitrary) way and use that but better (less arbitrary) if you have some values to base it on (i.e. perhaps the literature says the animal avoids slopes over x%).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.