Pollination ESS & yield

Dear InVEST pollination ESS model users
I have some difficulties to interprete the final results related to the yield index. pdep_y_w: Index of the total yield attributale to wild pollinators. What does this value exactly means? I used the villamette sample data. The value is between 0.014472 and 0.074658. So form 100% yield at a specific site a percentage of 0.01 is attributable to wild pollinators? As we did not include aboslute yield what does do model takes for 100% yield? I also checked the webinar, howver, to part explaining output/results.
Second the pdep_y_w map is quite strange, just some black squares? So only some pixels habe been calculated?
Third I suppose to verificate the model output with absolute yield data. However, I realized that in the webinar observed pollen limitation was taken. So is there a standardized indicator for validation?
Kind regards
Sibylle

Hi @sibylles ,

Thanks for writing in to the forum with your questions.

As stated in the User Guide, pdep_y_w is an “index of potential pollination dependent yield attributable to wild pollinators”. As an index, the best way to interpret values are as RELATIVE to one another, not as absolute values. However, if you’re insistent on an absolute interpretation, a value of 0.01 would represent 1% attributable to wild pollinators, not 0.01%. The range of valid results for an index is from 0 to 1.

Second, do not fully rely on your GIS’ symbology to interpret raster values! Black squares have no meaning on their own, but represent user defined or preassigned values by your GIS’ symbology. The most accurate way to read raster statistics is likely programmatically, such as using Python or R. Here are some tutorial videos on symbolizing rasters in ArcGIS and QGIS.

Third, I’m not clear, but I believe you’re asking about validation, which would require model calibration. I’m not aware of anyone using this model to get at absolute results, as it’s really intended for revealing relative patterns across space. But perhaps this section of the User Guide might help.

-Jesse

1 Like

Dear Jesse

Thanks lot.

Pdep_y_w. Yes I am aware that model is index based with relative values. I just wondered relative to what? But now it is clear, relative to the other pixel and not relative to the total yield. However the definition is “proportion of total crop yield attributable to wild pollinators”.

I adapted min and max to 0-1, but I just get the results for some pixels (see attachment)`?

Yes I am still struggling around how I should validate the yield indices (including the pollination dependent crops). Perennes et al. 2021 just validated the pollination potential but not the pollination service potential. I checked the tutorial and you compared x = predicted yield (fanegas/ha) and y= observed pollen limitation.

Kind regards

Sibylle

Dear Jesse,

Thanks a lot.
Any additional comments or answers to my follow up questions?
Kind regards
Sibylle

Hi @sibylles ,

As stated in the User Guide, " total_pollinator_yield_[Suffix].tif… [is only] generated if a farm vector is provided… Per-pixel total pollinator yield index for pixels that overlap farms, including wild and managed pollinators." For pixels that do not overlap a farm vector input, total_pollinator_yield results are not produced.

Results validation is best done by comparison with observed data from your region of interest, but I do not know of that ever being attempted for this model. Remember that InVEST is intended for revealing relative patterns across space.

-Jesse