But there are questions of scale. Diversity over what scale? In a fragmented landscape, diversity within a patch is not always best, as then you might not be able to support a population of x species. Take an insect with a specific foodplant. In a patch size of 1 ha it might take a dominance of the foodplant to sustain a viable population of a butterfly. If you increase the diversity, you spread the resources per unit area and limit the foodplant, but the unit area is finite and often acute. So the population of the foodplant may fall below the threshold required by the insect in that fragmented patch.
In which case the insect becomes locally extinct in that area, diversity is reduced, certain species become more dominant, the 'natural' balance is negatively affected. I don't know if it's intentional but you keep using the words 'you' and 'we', as if you feel we should be managing these habitats (all habitats/areas?) specifically for certain species. In some cases, yes, the bees you mentioned for example. But we don't know in all cases what the most important species are - the keystone species whose abundance is altered and will have the largest effect on its range of related species. Again it comes back to the question of which species do we concentrate on 'conserving', and which we don't, and our reasons for choosing these.
Even primeval forests are not a uniform pepper-like mix. You got different species dominating in different places, due to even small variations in soil, wetness, elevation etc. So species and diversity is not uniformly spread across a habitat. Woodland isn't just woodland. It's many different types of woodland, even within a wood.
Exactly - a diverse range of habitats and environmental variables supporting a diverse range of species.
Diversity aint the be all and end all. You can have 2 areas with the same species diversity but the compositon is very different. For example, an area covered with 5 plant species composed of 96% one species and 1% four species has the same species diversity as an area composed of 20% each species.
Diversity across a landscape, regionally or globally is obviously more 'healthy' (if you like) than a *pure* monoculture. But then you could argue that in modified Europe, a boar-free Britain that allows certain species (Bluebells?) to flourish adds to regional compositional diversity. Making England more like Germany is actually taking it towards a more monotonous pan-European woodland habitat by unifying composition.