Welcome to the forum @Allen!
As noted in the User Guide Interpreting Results section, effective_retention.tif is calculated using equation 48. This equation takes into account the efficiency for each LULC type given in the biophysical table, adjusting it based on whether there is more of the same LULC type downslope, and the critical length parameter for that LULC type, also in the biophysical table.
But it does not take into account how much nutrient is actually moving downslope to be retained by each pixel. What we’d probably want to quantify as the nutrient retention “service” is the actual amount of nutrient retained by the pixel, based on what runs off from upslope plus the efficiency of the LULC type. But the model does not calculate this. (I’m not sure why, it’s definitely something that should be improved, especially since the SDR model is similar and has explicit service outputs).
As for the calculation of retention services, you might be interested in this forum thread that contains a lot of detail about the NDR model, and toward the end they talk about retention, where the model designer really just confirms the recommendation in the User Guide.
I’d be interested to hear from people who have used this model more than I have!
~ Stacie