Good science vs reality

​​How science is supposed to work:

1. Observe something interesting about the world
2. Come up with a reasonable explanation for that phenomenon
3. Design an experiment to test that explanation

     According to this worldview, you should come up with question and hypothesis and then choose a system and a sampling scheme that best allows you to answer the question.

How science often works:

1. Collect as many samples as you can from as many places as you can
2. Figure out a question that you can (sort of) answer given your samples
3. Get really frustrated that your samples don’t actually answer your question

     In wildlife and conservation research, we rarely have the luxury of designing an experiment with nice sample sizes and rational spatial sampling. The organisms we study are usually rare and difficult to sample, so we have to take what we can get. For example, I’m currently working on a state-wide genetic assessment of bighorn sheep in Wyoming. The goal is to use genetic markers such as SNPs and microsatellites to identify populations and characterize the genetic diversity in Wyoming’s management units. This is an awesome project and I’m really excited about it. I love working with management agencies and the public. We’ve been able to get hundreds of samples from legally hunted animals and planned captures (like the tissue and blood samples in the picture).

​     However, the quality of the samples and where the samples are from make my life difficult when I’m trying to get 20-30 unrelated, mostly female individuals from each potential genetic cluster. How much can I really say about population genetics when I only have 3 samples from one drainage and 8 samples from another and most of the samples are male and half of the samples are moldy? And this isn’t even considering whether the genetic methods we’re using will be able to detect differences, because the rate of accumulation of mutations in a population may be slower than the rate of the processes differentiating the populations. It seems like then that I should adjust my question, which is troublesome because you shouldn’t fit your questions to your data but rather gather your data to answer your question!
​
So in the match-up of hypothesis-driven research vs opportunistic sampling, go for hypothesis-driven but be prepared to spend a lot of time considering whether your data can actually answer your questions.

---