Ramblings about what I encounter within the realm of the geosciences, as well as the occasional rant about nonsense.

01 August 2008

Correlation = Extinction

While people hopefully enjoyed a trip up to Nunavut, I figured I'd get back to talking about something tangential to my thesis. Extinctions.

I have talked about my almost school-girl like crush on the Signor-Lipps 1982 paper a lot, so I figure I will address another problem I view as systemic in extinction studies. The logical fallacy that correlations infer causations.

Perusing over Science Daily can show you how often a new study will shed light on an extinction claim. The problem with practically all of these studies is that they don't study an extinction themselves. Rather, they just study whether a possible trigger happened at about the right time.
Examples of this include:

Volcanic Eruptions May Have Wiped Out Ocean Life 94 Maa

Is This What Killed the Dinosaurs? New Evidence Supports Volcanic Eruption Theory

Asteroid or Comet Triggered Death of Most Species 250 Maa

Cosmic Impacts Implicated in Both Rise and Fall of Dinosaurs

When Earth Turned Bad: New Evidence Supports Terrestrial Cause of End-Permian Mass Extinction

These are just the press releases (sorry) so I am certain some of the claims are exaggerated. I am just trying to convey how often a new claim is made that ties an event to an extinction. Upon reading what the study actually covers, it presents no evidence that shows it is the trigger.

The closest you will ever see is something akin to this:
According to their research, the eruptions preceded the mass extinction by a geological blink of the eye.
In other words, this event happened at approximately the same time as we think the extinction happened. It is within the realm of possibility that something like this could cause an extinction. Therefore, the extinction was caused by this event.

Extinctions are the result of complex interactions within the ecosystem (among the biota, climate, landscape, etc.). Mass Extinctions kick it up a notch. Instead of worrying about a single ecosystem, you now are addressing all ecosystems. And any study that claims to show how an extinction happened needs to address how the perceived cause affected the entire ecosystem. To top it all off, evidence is a must. If I were to claim that a bolide caused the extinction of the dinosaurs, here is a (rudimentary and incomplete) grocery list of what I need to demonstrate:

  • An impact occurred at the approximately the right time
    • Alvarez et al, 1980 show this nicely.
  • No comparable impacts preceded it during the reign of the dinosaurs (otherwise, why did they survive the first one and not the last one?)
    • There is a problem with this, Keller's recent work raises this very question.
  • The dinosaurs went extinct after the impact, not before.
    • The problem here is the Signor-Lipps effect (sigh, it's dreamy) and the limits of biostratigraphy.
  • The effect of the extinction has to be consistent with the biota
    • More or less shown, large animals seem to have all been hit hard, but.....
  • The effects of the impact must also be accounted for on other aspects of the ecosystem
    • Possible results include: global fires, decreased primary productivity (photosynthesis), and acid rain. Why is acid rain important? Amphibians don't do so well with acid rain, yet they made it through the KP relatively intact. This hasn't been explained.
  • Potential other causes/inputs must be eliminated or accounted for.
    • Unfortunately, other mechanisms propose similar results. The Deccan Traps, for instance. But what about the lowering of sea-level and the mixing of formerly separated populations? (no geologic evidence would be preserved that would shed light on this, but this would have some influence on population dynamics).
  • Needs to explain why not everything died.
    • with the exception of the acid rain and the phibs, people have studied this one alright. Mammals are small, birds are small, ecosystems based on decay of vegetation provide a buffer.... (Though, in all honesty, these studies can basically be plugged into any of the potential causes. Meaning, these don't add weight to the bolide hypothesis).
  • Needs to explain the limits of this event.
    • did this event have anything to do with the marine extinction at the same time? Or is that an unrelated event? Here's a head-scratcher, did the two extinctions even happen at the same time or did the marine extinction start before the terrestrial one?

5/8 problems still need to be addressed before a tentative conclusion of the KP extinction can be reached. But extinction hypotheses routinely are treated like a done deal. I don't know what the reason is, but scientists really need to re-evaluate the utility of jumping the gun in terms of extinction studies. It is just bad science.


BrianR said...

very nice post ... the press release titles for ScienceDaily are getting increasingly bad in my opinion ... the press releases are usually okay, but sometimes it seems very apparent that someone else (an editor) writes the title to try and attract readers

Bryan said...

Thank you,

I wasn't trying to pick on Science Daily, they just happened to be what I was reading when I decided it was time to address what I perceive as a large problem with an area of study I have been working on.

Silver Fox said...

Thanks for pointing this out. I'm probably guilty of reading some of these extinction correlation papers or conclusions incorrectly, and hadn't really realized it!


All the Latin on this page is from my vague recollections from High School. There are mistakes in the text. I just was trying to get the point across

Between Los Alamos,NM and White Rock, NM

Between Los Alamos,NM and White Rock, NM
The photo of the travertine spring was taken in the small opening in the center of the image.

Lectio Liber