Here is one example of that tradeoff: "Scale-dependent effectiveness of on-field vs. off-field agri-environmental measures for wild bees" - https://www.sciencedirect.com/science/article/pii/S143917912...
Different species and crops may yield different results.
That is a big debate in conservation biology (known as "land sparing vs land sharing"). One of the major problems of the land sparing approach, which you outline above, is that there is nothing that makes (at least European) farmers as upset as the requirement to give up farming on part of their land. So while it's neat in principle, many conservationists doubt its political practicability.
So arguably the evidence we have is that if you consider only things that farmers like to be politically practical, then we're likely to see wildlife declines continue indefinitely.
That is of course why many people are looking for solutions that don't depend on the consent of the agricultural industry. One approach is to go over their heads and appeal directly to government e.g. the biodiversity COP agreement requiring 30% of land to be set aside for nature. Another is to hope that technological breakthroughs (e.g. precision fermentation) can fundamentally alter consumer demand for food products in a way that makes them much less land intensive (primarily by cutting out meat and other animal products, which use a disproportionate amount of land when farmed extensively, and produces a disproportionate amount of pollution when farmed intensively).
Of course, in the short term, the practical answer is "do both, wherever possible". Conservation organisations should be encouraged to purchase land so that it can be managed directly for wildlife. Farmers should be encouraged to adopt wildlife-friendly practices as much as possible. Governments should be encouraged to consider wildlife conservation a goal in itself when designing incentive schemes for the agriculture industry, and the taxpayers who directly fund much agriculture through subsidies should be encouraged to hold them accountable for their progress in this area.
Easier said than done, of course, but as the original article makes clear, the status quo is chronic failure, and there's no reason to suppose that can be changed within the current paradigm.