I'm going to summarize, likely poorly, my understanding of the presence of antibodies, what that means for the immune response, and why the decreasing number of those with antibodies may not necessarily be alarming.
First, here's a graph, no idea or not if it will show up but you can click the link to see the graph.
Basically, the main takeaway is the IgG antibodies were known some time ago to fade after the infection was beaten down by the body. It may now be better known that those IgG antibodies last only 6 months or 3 months, but likely not a year. So, that's better known...which is okay.
Antibodies more/less attach to a virus and prevent it from getting into the cell where the virus replicates. I don't know if there is a good analogy here. But, let's go with ...
[analogy] a virus (sars-cov-2 in this case) is a home intruder. An antibody (IgG) is a guard dog. Ideally, the intruder tries to come into your house, but the dog sees him, latches onto his arm in the front yard, and the intruder doesn't come inside the house.
But, at first because you live in a nice area and intruders are a novel thing in your neighborhood, your dog is simply Hercules and not yet The Beast. He isn't a trained assassin and is used to idly sleeping and drooling on baseballs that wonder into the backyard every now and then. He isn't trained to detect intruders and he isn't living outside. However, fool me once shame on you, fool me twice shame on me; and you get broken into. At which point, over the next couple weeks you train the dog to become aggressive towards inruders and put him outside. You have to teach him to be The Beast. OVer the next couple weeks, that dog is a killing machine and at watch 24/7. As a matter of fact, you create a couple thousand clones of The Beast to put in the front yard. But, with time and no intruders around, the dogs become lazy. Some of them run away. Your yard starts to smell and complaints are filed with the City. The City comes by and says, yo, you can't have this many dogs so you find them nice homes down the street. But, of course, you're going to keep Hercules plus a couple and even if they return to their previous life as inside dogs and don't stand at watch 24/7, they haven't forgotten about the threat of an intruder. If one so happens to come around and tries to steal a Babe Ruth signed baseball, even if the dogs are back to being Hercules, they are ready to turn back into The Beast in a heartbeat.[/analogy]
I don't know if that analogy is perfect yet (though antibodies certainly don't have to be actively detectable to be ready to ramp up production), nor does the imperial college that released the study. They just note that the amount of guard dogs in the front yard decreased. The body still has memory cells and its entirely possible those memory cells, while dormant, aren't making IgG but stand at the ready with the first detection of Sars-Cov-2 to kick back into gear to ramp up IgG production instantaneously as opposed to the week or two it originally took during the first infection. IF that's the case, while immunity may not be 100%, it's a pretty good bet even without detectable levels of IgG, any long term effects and/or fatalities will be greatly diminished the second/third/fourth/etc infection.
And, that would be rosy. Of course, we also can't be certain yet that the virus won't mutate to the point where even if IgG is made, it will be effective. Or, worse, the virus could mutate to use IgG against the host making it more dangerous.... which would be akin to The Beast returning and eating James Earl Jones.
All that said, it's not as cut and dried as it is made out to be. One thing is certain, and that is mutations of SC2 are happening far less quickly than they do with the flu. So, the real question is rather or not the memory cells will react to produce IgG or other antibodies capable of fighting of SC2, how effective will they be if so, and how long will that memory last if proven to be effective?
