Im interested in how we can determine what are the optimal conditions for crops and how deviations from these optimal conditions affect yield. In the literature I have found two main approaches and I was interested if there were more ways that are used or if anyone can think of a more creative way to test the link.
Use a greenhouse experiment to isolate all variables bar one and vary it. Measure the change in yield based on that variable.
But this approach is time consuming and it is difficult to include interactions.
Use a model that relies on some base knowledge of that plant - i.e. the base temperature at which growth is zero, and a rough idea of optimal ranges. But requires this base knowledge already.
Potential Approach Three
Use multiple linear regression in combination with field data to create a linear or polynomial model of each variables effect on yield. Interactions can be included easily. But relationships are unlikely to be linear and without prior knowledge it is hard to decide on non-linear models.
Can anyone think of other ways that could be used to test these links?