Change. One thing we can depend on. Especially in nature. It seems that sometimes the best way to mimic nature in an experiment is to leave it to nature. As opposed to field plots where you control diversity, in natural systems, you have to embrace diversity, change, and yes, stochasticity.
These can and should be seen as factors inherent to natural processes. In fact, if nature were dependable, fair and balanced (that's
actually fair and balanced... no homage to Fox News here), the world would look very different. For instance, one ecological hypothesis holds that a mild-to-moderate level of disturbance is a factor that maintains diversity in communities. That is, if several plants are competing in a benign, nutrient-rich, favorable environment, the superior competitor for resources will win, hence, excluding the others. Events like, say, a big rainstorm, a flood, high winds, animal foot traffic can serve to keep all of the species on their toes, and ensure continued opportunities for all of the plants to have a fighting chance.
For instance, my dog goes for a lovely jaunt, chasing a balloon through a field with only one species of flower.

The other plants have been out-competed, but their seeds are waiting in the soil to germinate.
This is a common route for my dog, however, and I find the following year that another species is making an inroads in the community, taking advantage of the trampling effects of increase

in soil compaction, and increased light, for instance. He continues to love to run here and eventually, this area is now converted to the white flowers. We move away, leaving the field to itself, and years later, we return, only to find yellow flowers abound once more.
Currently I'm dealing with a problem of variability of resources in nature. Not a disturbance process, but a stochastic process. This process may be the key to the establishment of a species. If all years had average rainfall, it is even doubtful that this species could gain a foothold in certain places. It seems that a higher-than-average year every now and then allows the species to establish, and then just survive the low years. Therefore, models including only year after year of average weather may miss something important.
Take, for instance, these three years of rainfall in southern California:



If I tried to simulate an average year, how exactly do I do that? It seems, variability is part of the system. Constancy is not. Not only does this make it very difficult to do a field experiment and get it done in 1 year, but the variation itself may be important.
As a foreign professor told me yesterday, "if you want to see what happens in the nature then you have to leave it to the nature." Nature... much love.
.