915iS High Manifold Temperature

All I am saying is that it is "Theoretically" possible to get another ~3% power for every ~10°C of temperature decrease.
Whether Rotax takes specific advatage of the option is a different question.
The engine is rated at a specific Airbox Temperature and most likely under sea level standard ambient conditions, which you will likely never see in real life.
The "iS" engines are known to operate "Lean" of optimum and Open Loop. (No Feedback)
Unlike Auto engines, there are No Exhaust Gas Oxygen Sensors to use to adjust the mixture on the fly.
They operate Lean in ECO mode and Rich in POWER Mode according to an internal table.
The ECU is conservative and does not try to push to the edge of the envelope.
Reliability is considered to be more important than performance.
The EGT probes are informational only and Not used by the ECU.

50°C might (guessing) be the point where the mixture crosses onto the Rich side, so lowering the temp will likely just tend to increase the Lean margin slightly.
Will this produce more power? Maybe.  Will this improve fuel economy?  Maybe.
A colder Airbox will certainly help to increase the 915's Critial Altitude above 15,000 ft.

The exact details are not published so all of this is really just an academic exercise.  
But it is always better to understand the underlying principals than just blindly follow the instructions.

Does that mean it won't run as efficiently as a carb engine where the pilot manually adjusts mixture to lean in flight? If you lean mixture until engine runs rough then enrich slowly until it smooths out, keeping an eye on EGT, will that result in a leaner and more efficient running engine than a Rotax EFI controlling mixture that blindly uses a default ECO map and takes into account no feedback from the running engine?

Comparing the 912iS to the 912ULS should help answer that question...  The injected engine produces the same power, but burns a good bit less fuel.  (And yes, I realize that neither allows the pilot to directly manage the leaning process, but you see the same pattern repeated in the O-360 vs IO-360, etc.  There's a reason all cars sold these days are fuel injected...)

Pilot Joe wrote:

Does that mean it won't run as efficiently as a carb engine where the pilot manually adjusts mixture to lean in flight? If you lean mixture until engine runs rough then enrich slowly until it smooths out, keeping an eye on EGT, will that result in a leaner and more efficient running engine than a Rotax EFI controlling mixture that blindly uses a default ECO map and takes into account no feedback from the running engine?

Even though the 915 may not have all of the bells and whistles to maximize efficiency that a modern automobile has, it does need to. Unlike an automobile, aircraft operate mostly in a relatively narrow power band, so the engine can just be optimized for that. The proof is in the pudding -- much higher efficiencies in terms of pounds of fuel per brake horsepower at cruise power settings.

James N Parker wrote:

Comparing the 912iS to the 912ULS should help answer that question...  The injected engine produces the same power, but burns a good bit less fuel.  (And yes, I realize that neither allows the pilot to directly manage the leaning process, but you see the same pattern repeated in the O-360 vs IO-360, etc.  There's a reason all cars sold these days are fuel injected...)

The injected engine also controls spark timing whereas the carb engine has fixed spark timing and optimized for WOT. How much loss of economy in the carb engine is due to that, and how much is due to the ECU running a more lean engine than a pilot could? I am skeptical that an open loop ECU could manage mixture more effectively than a pilot but maybe I am overestimating the difference between the two. I'm reading that in eco mode the 915iS runs with a lean mixture of lambda 1.05 (power mode runs at lambda 0.88). That is in theory but it'd need a closed loop system with an oxygen sensor to keep it accurately at 1.05. Do Rotax engines just not need closed loop because the way they are run lets it stay at close to 1.05 without feedback?