Pollination: crop type without farms map

Dear community
I have a question related to the InVEST crop pollination model. Is it possible to integrate pollination dependent crops without using a farms map (similar to Perennes et al. 2021)?
Reason: I have a map of pollination dependent crops, but not the floral resources and nesting suitaiblity for the farms / pixels, so I would prefer to use the guild and biophysical table.
Kind regards

Hi @sibylles -

You need floral resources and nesting suitability for all land use/land cover classes in order to run the basic model. The land use/land cover map generally covers your whole area of interest, including the farms. Running the basic model, without the optional farms map, will still produce a map showing abundance of wild pollinators on each pixel, including farms. But you won’t be able to incorporate any of the other farm-specific information that is needed to calculate the crop yield index due to wild pollinators.

If you do have the other farm-specific parameters (crop_type, season, p_dep and p_managed), and want to assume that only wild pollinators are used (no managed pollinators), you could try using the farm map, setting the on-farm floral and nesting resources to 0.

However, I’m not sure how doing that interacts with the model calculating abundance on the underlying LULC map. It seems like pollinator abundance is first calculated across the whole LULC map, which includes pixels within the farms. But then there’s an additional calculation that takes into account the farm-specific definition of floral and nesting resources. Do the farm-specific values replace the general LULC-based values in the pixels where the farm is located? Or are they in addition to them? I’m having trouble explaining my confusion clearly, but perhaps someone in the software team understands and can clarify?

~ Stacie

Dear Stacy

Thanks a lot. Yes I am also confused a little bit. I have everything for whole Switzerland

  • LULC map
  • Biophysical table
  • Guild table
  • Crop type (even differentiated for different pollination dependent crops) per 25 m pixel (but not per farm)

However not the other parameters from the farms map. I do not understand why we need fr and n two times.


Dear all,
Has anybody tried to run the crop pollination model without farms map but with crop type? As mentioned I have LULC, biophysical table and guild table.

Since the farm map is optional, do you get the information that you need by running only the first part of the model? It will create maps of pollinator supply based on the resources in each LULC class. So if you have information that differentiates each crop type in the LULC, you should see differences in supply to each crop type.

~ Stacie

Dear Stacie

  • I suppose without farm’s map just average values per crop type. However I have a pixel based map of crop types. Aim: see also regional (spatial) differences.
  • the second aim is to receive the per pixel pollinator yield index. For that I think I need the farm’s map. However, instead of farms I have pixels. How should I produce fr and n per pixel? Is this automatically produced. I still do not understand the overwriting.

Kind regards

Sibylle, have you looked at the description of the model outputs, or tried running the sample data to see what the results look like? The non-farm output from the model provides per-pixel values of pollinator abundance.

If you actually want the model to calculate a yield index (taking into account managed pollinators etc that are associated with farms), the way the model works you do need to provide a polygon farm map and required model values. You could try extracting only the crop classes from your LULC raster map and convert them to polygons for use as a farms vector input. This is most likely to create larger polygons of farms, not individual square polygons representing pixels though, I’m not sure how you’d do that, and you’d end up with a lot of polygons.

The only other thing I can think of is to consider doing the on-farm calculations yourself at the pixel level, although you’d have to think about whether the equations given in the User Guide are appropriate at the pixel level, and/or how to adjust them.

Sorry I don’t have anything more productive to add. If anyone else does, please do!

~ Stacie