No, this isn't impossible. It's been done in a number of paleoclimatic/paleobotanic studies. The rate of ecological change can be measured against the rate of climate (temperature) change using fairly direct measures of temperature, such as stable carbon or stable oxygen isotope ratios.
As you note, I think he goes out on a limb labeling the changes as both abnormal AND the result (at least partially) of human influence, but it is possible to compare current rates of ecological change to the rates of change that have occurred in the past.
It's hard to know what the sample size would be, because it is hard to know what the total number of living organisms is. However, let's say we decide that we need a sample size of 1,000.
You would need to probably break that down in to a certain number of ecological climates, and then take a random sampling in each. Let's say we broke it down in to 10 climate zones, and we took a sample of 100 organisms from each.
How exactly would you select random samples?
Do you really think that this guy did that?