Something isn't right. As I increase the DC-AC ratio from one, production goes UP until about 1.2, then it starts going down
Plausibly correct if inverter idle losses are being considered; but you will need to dig into the actual equations to confirm this.
The PVWatts technical reference is
https://www.nrel.gov/docs/fy14osti/62641.pdf
The inverter efficiency model used is shown on page 15. The model isn't aware of specific products at specific power ratings; you enter your DC kW rating and your DC:AC ratio, and that defines the 100% rating of the inverter. Then the modeled DC output of the array is compared to that 100% rating, and the inverter efficiency scaling is applied.
The inverter efficiency model curve on page 15 is efficiency (%) vs per unit power. Re-graph it as efficiency (%) vs power (kW). Decreasing the DC:AC ratio has the effect of shifting this curve to the right, so that more of your actual production gets lost to the low efficiency region when the inverter is operating at low % power.
That is just me hand-waving an explanation. It would be nice to see intermediate results of the PVWatts model, so that you could see 'this much production was lost to clipping' 'that much production was lost to inverter efficiency', etc.