Not 100% sure on this, but some thoughts. "The two small floating circles represent pond returns" might be the key as they represent connections (head connections) to your pond. As your pond probably has a large surface area the level there may be fairly stable for the most part. But for sure, any change in the pond level (head) will reflect back on your system as shown and affect flows. Check
https://www.engineeringtoolbox.com/p...ial-d_636.html re pumps in parallel. You are operating at point three in the graph. Any change in the pond level will reflect back on the pump head, affecting flows. The issue is time. Flows in m3/hr (or whatever units) translate to pond level rise or fall in a complex time based formula. Steady state, after time has passed things should stabilize you would think. But not always, often because we are too impatient. I have a complex pond arrangement. When there is heavy rain, level goes up. I dump water to a storage tank. If the quantity of water dumped = quantity of water caught by my pond/river system, then perfection: All levels should equalize. But I seldom get it right and paradoxically when there is heavy rain, my RDF 'fails' and I pump out water leaving my system at low levels.
It's late here, it's Friday night. I've had a few beers, a couple of glasses of wine and a Black Label thinking about this. My feeling is that you have a complex arrangement (like many of us I guess) and the phenomena you see is a complex flow/level/TIME related issue. We all understand flow and levels, and that levels can affect flows. But we never take into account the TIME it takes for flows to affect levels. Not sure if this helps.....
Meanwhile, I have a tropical storm outside. Off to dump some (but not too much) water to storage.
Dan