So there are at least 3 studies performed by different teams in this article that indicates you're incorrect in your assertions as applied in practice.
That article is ... confused. Donald Davis, the main source for most of the article, suggests that a lack of selection towards nutritious crops as the cause of lower nutrition, while the article itself claims that the cause is poorer soil quality. Noticeably, neither of these points support the argument that engineered crops are any worse for the soil than past crops.
Also, the article suggests that the studies themselves are comparing current data to data generated 20-50 years earlier. Comparing two modern day experiments to each other can be quite difficult, as even small differences can have huge ramifications in the results, even if the protocols are supposedly the same (I personally saw this play a few times in grad school). For example, quality differences in the filter paper used can result in a much higher retention of mass in one experiment compared to another. Without a body of research measuring the comparability of the compared results (please link if you find it), the comparisons cited in the article are not very convincing.
Actually, if you can find the original studies cited in the article, I'd love to see them. The ideal way of measuring this would be to take a heritage strain, and grow it along side a modern strain the the same soil. That would at least help figure out if modern crops really have selected against nutrition.
Conjecture with me for a moment, friend: if given fruit X yields 100% nutrient fruits, is then selected for volume, and in the increase of size, the proportion of nutrients declines it follows, to me, that the increased volume is creating a diluting effect. To me it seems as if any increase of proportion in volume had ought to scale right alongside the proportion of nutrient values, unless that compromises the organism. This is the framework I'll be working from.
Several factors come immediately to mind, is the plant reaching a threshold wherein micronutrients are at "unity" and so uptake is reduced and thus reaches a plateau (or saturation), and so distribution into fruit is proportionally diminished as volume increases? Is there some natural law that is prohibitive, something like the square-cube law? Or are the plants locally depleting the nutrients, and then relying on natural diffusion? Is there a "long-range dependency" that is opaque? All else failing, I would suggest that it's highly probable that the nutrient disparities that are evident in these studies could reflect that maintenance of the proportion of nutrients to that of volume might leave the plants fated to death, and so our selection process is predetermined to either volume or nutrient content. Which begs the question, at what intersection do we find the highest degree of efficacy in selecting foodstuffs?
Given a plant that could expand both volume and nutrient proportionally, what would the result be? I'd conjecture rapid depletion of locally available nutrients, up to the point where diffusion and natural deposition becomes an ineffectual mode of conveyance and demands manual upkeep else the organism would be given to death.
Perhaps I'm presumptuous, I don't mind being so, I'll have to ask you to forgive me for my ignorance. I only ask for your participation in this discussion as it seems you're privileged to have a mind much more discerning and honed in this craft than mine own.
Plants are ~70-90% water, and of the dry mass, 70-95% is carbon, the structure of the plant which is extracted from the air (splitting the C from the O2).... the remaining nutrients are from the soil + dissolved in rain water, although plants do absorb a limited amount of nutrients through their leaves via dust etc.
Also, the article suggests that the studies themselves are comparing current data to data generated 20-50 years earlier. Comparing two modern day experiments to each other can be quite difficult, as even small differences can have huge ramifications in the results, even if the protocols are supposedly the same (I personally saw this play a few times in grad school). For example, quality differences in the filter paper used can result in a much higher retention of mass in one experiment compared to another. Without a body of research measuring the comparability of the compared results (please link if you find it), the comparisons cited in the article are not very convincing.
Actually, if you can find the original studies cited in the article, I'd love to see them. The ideal way of measuring this would be to take a heritage strain, and grow it along side a modern strain the the same soil. That would at least help figure out if modern crops really have selected against nutrition.