i suggest a simple ranked index (R) based on sediment yield and distance from sub watershed to main watershed outlet.

R= RY + RD

R: aggregated ranked index. RY: Rank of sediment yield. RD: rank of distance to man outlet. Both from 1 to N. As the lower value, the more  importance of subwatershed.

Y= Ys/Yw ; both in Mg/ha

Ys: sediment yield of each subwatershed

Yw: sediment yield for the whole watershed.

Then rank Y from highest to lowest values to obtain RY for each subwatershed.

D = Ds/Dw

Ds: distance from subwatershed to watershed outlet.

Dw: Maximun subwatershed distance from subwatershed to main outlet in watershed.

Then rank  D from highest to lowest values to obtain RD for each subwatershed.

Finally, obtain RY + RD for each watershed. The lower values implies the more importance of each subwatershed.

The ideas above are not bad.  I really don't know of the perfect answer.  Roehl  (1964 I think) estimated sediment delivery by watershed size for differing physiographic areas using dam efficiency as an indicator.  In general, his data supports smaller hydrologic units deliver more sediment per unit area than larger units.  The term subwatershed does not mean the he same to everyone, but by that term, I assume you are probably 10-40 thousand acres, maybe 4th to 6th order streams.  If you know or are aware of the Rosgen stream types, braided streams D type are efficient at retaining sediments, and other stream types such as G and F types are normally unstable, and more efficient at both eroding and delivering sediments.  Check out wildlandhydrology.com.  However if the ultimate goal is to reduce sediment, knowing streams or sections that are sediment sinks, aggrading, etc., those may be your friends, even though habitat for fish may be bad.  Streams that are degrading or actively widening to reach stability, those are important to find, along with major sediment sources along roads, landslides, gullies, etc.  Habitat for fish is not that great for these either.  The middle of the road are streams in balance as they have regular access to their floodplain at bankfull flow so their bank erosion is limited, and they are able to store some sediment during over bank flows during floods.  By identifying your most critical problems by stream reach or specific activity, such as roads or urban areas, you can more effectively target your objective.  I am not very familiar with SWAT, but it may include some of the land us and other factors that I would consider.I use RUSLE approach is that I have developed and adjusted for our specific circumstances, vicinity land uses, road types, etc. over the years.  Over the years, I have had mixed feelings about beavers, but I know in many areas of the USA, they were native species, and the early explorers suggest clear and clean waters.  I have seen sediment accumulation, and beaver dam failure.  My guess in lower gradient braided systems with ample rain and rapid vegetation response from beaver eating of riparian vegetation, beavers may be beneficial if sediment reduction is critical.  However, they can also stop up culverts or do other damage if they locate in a the wrong places.  Treatments need to fit the circumstances, and you have the right idea in looking for ways to prioritize your effort or advice.