There's a diagram of the structure of human Lp(a) here. The apolipoprotein(a) molecule is the curvy bit partly wrapped around the lipid particle. The section which varies on a genetic basis is the number of repeats of the kringle IV type 2 (shown in black). There can be as few as 10 repeats or more than 50. The rule of thumb, within a given population, is that the lower the number of repeats there are, the higher your total plasma Lp(a) is likely to be. NB comparisons between populations are notoriously difficult. Better control your variables and stick to one population.
It almost looks as if the liver wants a certain number kringle IV repeats in the circulation. If there aren't many kringle IVs in the gene, the liver puts more whole protein molecules out to get the number up. Each protein is attached to a single LDL cholesterol particle. Short repeats mean more individual apo(a) proteins are needed to get the number up, which means a higher Lp(a) concentration. It's not a complete equalisation of kringle IV repeats, but that's the general organisation.
The vegetarian Bantu have two things which are special about their Lp(a).
The first is that, despite considerable intermarriage with fishing Bantu, they are genetically short on the kringle IV repeat front so, naturally, they produce more apo(a), attach each to an LDL, so have higher Lp(a) levels across the board compared to their more carnivorous cousins.
The second is that, if you control for this effect by selecting and comparing two genetically apo(a) matched groups, one veggie and the other fish eating, the difference decreases but is still significant.
What can you make of this? The first thing must be that there is a selection pressure to maintain the differences in apo(a) genetics. Either the fishermen don't need much Lp(a) or the vegetarians need lots. I think the vegetarian environment is such that people with high Lp(a) are more likely to survive.
Second follow on is that the extra Lp(a) from the genetics of the vegetarians is not enough. There is some scope for "pushing" Lp(a) levels up or down from the level you might expect from the genetics of apo(a). The vegetarian environment requires higher Lp(a) than the fishing environment.
We know from intervention studies such as DELTA that a simple change of 7% of calories in the SAD from saturated fat to carbohydrate will increase Lp(a) levels by 20%. What does this mean?
If you subscribe to the "Lp(a) is a suicide lipoprotein" theory, you have to conclude that replacing saturated fat with carbohydrate makes your liver want you dead, by 20% more than it did when you ate butter.
Using monounsaturates instead of saturated fats is not quite as bad, the hepatic homicide lipoprotein only goes up by 11%.
The other way of looking at the changes is to suggest that your liver either monitors your macronutrient intake (you think it doesn't?) or some marker of vascular damage. If you do something which is either is outright damaging, or which is sensed as potentially damaging, your liver acts to save your life. If it "perceives" that you need more Lp(a), you get it.
What each of us gets in terms of the genetic number of kringle IV repeats is probably determined by where our personal ancestors lived. Not much we can do about that. What ever we do to tweak this background level of Lp(a) production is up to us.
From the DELTA study it's pretty clear that no one is going to drop their Lp(a) by eating "healthy" monounsaturates or carbohydrate. From the fish oil study we can see that you won't get any joy from adding omega 3 fatty acids either. I've not seen the effect of omega 6s on Lp(a), but I'd guess they're probably as bad as carbohydrate.
As I see it Lp(a) is a very interesting lipid. I think it's hard to get much information from a single measurement in isolation, but changes in Lp(a) probably give you marks out of 10 for your changes in food choices. "Healthy" oils and carbohydrate score you zero. These are likely to damage your vascular system as judged by your liver's increased output of Lp(a). A surrogate for vascular damage would be the increasing blood pressure with age seen in the vegetarian Bantu, probably related to their 82% carbohydrate diet, and so they will need more Lp(a). They get it.
Down at 70% carbs the fish eating Bantu cope well and don't have degenerating arteries to raise their blood pressure, so they don't need so much Lp(a). They don't make it.
I'd predict that substituting beef dripping for both carbohydrate and olive oil would give you the lowest Lp(a) concentration within your genetic window.