With all due respect (since I do not really know what is going on in Jeff's engine) I think this particular data (which was hard to read in the link) for fuel injection engines is misleading as to this issue (why does the Floscan say Jeff is using more fuel at a lower rpm with a fixed throttle). When an engine slows down with a fixed throttle plate, since the engine does not "suck" as hard, the manifold pressure rises. That, along with the fact that the intake valve is open for a longer time results in more air flow into each stroke. This is important information to fuel injected engines since the ECU must calculate how much fuel to shoot in. More air means it should inject more fuel for each stroke to keep the ratio correct. You can see why calculations for fuel injected engines spend a lot of time trying to determine the volumetric efficiency (how much air is getting into each stroke). It is the base datum for maintaining fuel to air ratio. But it doesn't mean much in terms of fuel usage per hour unless you multiply by rpm and displacement.
Carburetors (such as Jeff's) supply fuel in relation to the air flowing though them (ideally) and don't worry about where the air fuel mixture is going. They only care about total air flow so unlike an injected engine they do not worry about how many cylinders they are feeding or how often the intake valves open. Of course it really is not that simple which is why we have power valves, accelerator pumps, etc. And maybe that stuff is Jeff's answer. They are there because the carb fuel flow is slow to respond to the input of changing throttle plate position or changing load (decreasing rpm and vacuum at fixed throttle position). Or maybe the Floscan data really is not that accurate in the short term. Don't really know, but I do agree that fuel injected engines (and in fact all engines) use more fuel and air per stoke at lower rpms for the same manifold pressure and at lower rpms the manifold pressure should be even higher, which means even more fuel and air per stroke. But I am not sure that is the answer to the riddle since there are less strokes per minute.
@Mark As for the last suggested observation, I guess your neutral at the dock and in gear underway example might be a good test. My belief, is that the same throttle plate position that gives 2500 rpm in neutral, might give perhaps 900 rpm under load and that fuel consumption would be less not more. (addendum..Or maybe it would be the same, see below..) One could put some stops on the throttle plates and see what happens under those two conditions. I would do it myself but I don't have the Floscans. But that test would sort of simulate Jeff's observation. Or you could get a reading underway at a steady rpm on a calm day and then drag some buckets and see what happens to speed and fuel consumption under the increased load.
Now if you want to get the same 2500 rpm under load as at the dock, then you would have to increase the throttle plate opening. Vacuum would be reduced compared to the dock, more air would be ingested for each stroke at 2500 rpm (because there is more absolute pressure in the intake manifold) and more fuel would go in along with the air. But that does not simulate Jeff's observation.
After thinking about it, I believe we should assume that all these systems (carburetor and injectors) do a good job of controlling fuel air mixture. So engine air consumption per hour is also a measure of fuel consumption. And air and fuel are a measure of power output. Air consumption (and therefore power output) for a given engine is purely a function of rpm and manifold pressure (absolute as in above zero). Manifold pressure is a function of throttle plate opening divided by rpm (we can assume a linear relation over the small rpm change in Jeff's case). If your write the equation for this, you find that rpm cancels out!! and it says that air consumption (and fuel consumption and therefore power) are purely related to throttle plate opening.
Now I know that we are talking about steady state and also at the ends of the rpm range other stuff comes in, but this suggests that at a given throttle opening if we see varying load (climbing up a wave) we will see an rpm change (boat slows down to reduce speed related drag so total load remains the same) but will not see a change in air or fuel consumption (because power output is the same). Of course marine engines seldom see a varying load unless throttles are opened to increase speed by producing more power. There are no hills on the water except for Jeff's waves. So I still do not see an explanation other than to observe that as load increased, rpm dropped but then manifold pressure increased so that air flow into the engine remained about (or exactly) the same. Therefore fuel consumption did not go down even with the reduced rpm. And the same amount of air would go though the carburetor. So I am reversing my previous position about airflow and fuel fuel and rpm, but still do not know why fuel consumption would go up (not stay the same) when he hit the waves.
Interesting subject, still unclear to me but I enjoy the discussion.