| [Thread Prev] | [Thread Next] | [Thread Index] | [Date Prev] | [Date Next] | [Date Index] |
pre-script: I accidentally sent this out in NeXT Mail format, which meant that many of you probably got this message earlier and couldn't make it out. Thanks to Andrew MacRae for salvaging it. This is quickly spinning out of control into a flame war. I apologize in advance for perpetuating it, but Norm has presented such interesting and substantive arguments that I feel compelled to respond. And now all of a sudden I find Una joining in as well. I would really like to see some more input from others, especially from people who actually _do_ work on the K-T (my interests are more focused on the Cenozoic itself). General comment on MacLeod: A lot of our disagreements seem to stem from a fundamental philosophical split. Norm seems to be taking a Popperian position: scientific hypotheses must be tested, and if they fail, they should be discarded (I'm exaggerating his position to make the distinction clear). My own view is more Kuhnian: if a hypothesis explains your data better than no hypothesis at all, it shouldn't be discarded until a _better_ hypothesis is found. With this in mind... "your proposed interpretation is based on a perceived failure to prove a negative statement... what positive evidence exists that corroborrates your hypothesis..." See above comment. "I'll stick with my interpretation of no Brazos planktic foraminiferal extinctions being associated with bolide impact or proved beyond reasonable doubt to result from bolide impact." Back to fig. 4A McLeod and Keller 1994 (sorry, I switched the authors by mistake in my last post). Norm excludes species 1-6 from the end-Maastrichtian event; they "fall short" by 110 cm in one case and 30-60 in the others. All of them are clearly rare. I have been told repeatedly by assorted invertebrate paleontologists that supposed pre-boundary rare foram extinctions in individual cores turn up at the boundary in other cores (only species 6 appears in the second core [ODP Site 738], and I can't find it in the 738 range-chart). Norm says he has done confidence interval analyses, but in this paper he hadn't gotten around to it yet. Just eyeballing the data, it seems difficult to exclude the K-T boundary from the upper confidence interval for these species. He grants 7-16 as K-T victims, but excludes 17-28 because they range into Zone PO (or P1a, based on two occurrences). PO is defined as "Danian" by four species (35-38) in the Brazos core. Three of them appear to be marked by a grand total of five specimens. If PO turns out to be a zone of reworking and the six rare species are actually K-T victims as I intimate, the boundary takes out 28/34 = 82% of the fauna. If not and all of Norm's qualifications hold, it still takes out 10/34 = 29% within a few thousand years. This is a mass extinction on _either_ account. If the nine "PO victims" (17-25) really do indicate a second event and the four Danian species (35-38) are _not_ coeval with them, that event takes out no more than 9/18 = 50% of the standing fauna. However, it takes out one and possibly seven fewer species. My calculations here go to prove two points. First, the K-T event was an out-and-out catastrophe assuming Norm's data and _interpretation_ of those data. It was as bad or worse than any possible secondary event, again on _his_ interpretation. Second, Norm's claim that "no Brazos planktic foraminiferal extinctions [are] associated with bolide impact or proved beyond reasonable doubt to result from bolide impact" - no extinctions, not even one - is badly contradicted by his _own_ data. I can't see any way around a _temporal_ association (this is all that we are debating here, not a causal relationship), _unless_ the impact level and/or the ranges are misidentified _by Norm_ in his own core. Or else, of course, unless Norm is taking a very strict Popperian position and considers _no_ scientific hypothesis ever to be "provable." I'm sorry if I'm putting this rudely. It's nothing personal, I'm just trying to make the issues clear. "The existence of such long term effects tells you nothing about the duration of the causal mechanisms." I agree, but the existence of very _short_ term effects _does_ tell you about the duration of causal mechanisms - namely, that those mechanisms must _also_ act on very short time scales. For me, an extinction on the scale of 10,000 years or less (e.g., the strictly-defined K-T boundary events discussed above) rules out virtually any causal factor that could be called "gradual." If a series of impacts, eruptions, sea level changes, etc. on _that time scale_ are the cause in question, then they are _catastrophic_ impacts, eruptions, etc. "Rosemary Askin's data, Kirk Johnson's data, and Gerta Keller's data all show increased survivorship in the high latitudes." Yes, and this _confirms_ the Chicxulub-based impact model because Chicxulub is in the _tropics_. I believe that all of those data sets _do_ show extinctions at the boundary - just not as many. Furthermore, as far as I can recall impact modelers have predicted very short-term _cooling_ events but not heating events. And finally, the long-term climate change at the boundary is towards cooler temperatures. All of these things _are_ explained by or fully compatible with the impact model. I would also like to mention the Raup/Jablonski analysis of molluscs, which showed no clear biogeographic signal but only _if_ you exclude the rudists, a diverse and strictly tropical group that went completely extinct at the boundary! "Dave Archibald's data on lower vertebrates from middle lattitudes does not corroborrate predictions of some post-impact atmospheric models." With all due respect to Dave, I strongly disagree with his interpretation of these data. I would prefer to argue this with him personally and not "behind his back," so I will simply comment that the taxa he analyzed are not taxonomically or taphonomically uniform enought to interpret differences in their extinction percentages; that the sample sizes involved are too small to allow such interpretations without testing for significance, which was not done; and that his "predictions" of the impact model are based on his own assumptions about what the physical impact sequelae should have been, and on the connection between the _physiology_ of Recent organisms and the impact survival probabilities of their frequently distant Maastrichtian relatives. I see Dave's results as a house of cards - interesting, but not strong evidence one way or another. "Hypothesis testing is a two-way street..." Once again, _what is the other model_? Major mid-latitude extinctions happened globally in almost every group that's been looked at. The impact model accounts for this; it is better than no explanation at all; so the only remaining question is whether a rival causal hypothesis does a _better_ explanatory job (_not_ a null hypothesis, because the null hypothesis has already been rejected over and over again). "...this is an instance where parsimony is in the eye of the beholder. In phylogenetic analysis parsimony is determined by the number of ad hoc hypotheses required to account for the observed data. Admittedly under a progressive extinction model a separate ad hoc hypothesis is required to account for each observed extinction. However, under the impact scenario this is also true." If I understand this correctly, what you are presuming here is that the evidence for secondary waves (or a secondary wave) of extinctions is incontrovertible. I don't agree. We can look at other data sets, but in yours I think it is far from clear that most or all the "PO victims" (18-28) are anything other than reworked K-T victims. If this is correct, there are only six Cretaceous species left to form a cohort of secondary victims. Once again, I am not a micropaleontologist, and I would like to hear from folks like Steve D'Hondt, Jim Zachos, or Marty Buzas who are more qualified to comment. Another very important point here is that _any_ catastrophic extinction will result in short-term increases in turnover rates. This is a simple and unavoidable consequence of the logistic growth model, which is clearly relevant here. It would be highly surprising _not_ to see a fair number of true and pseudo-extinctions in the middle-term wake of a catastrophic mass extinction caused by _anything_. Even if all the last two points can be dismissed as flame-bait, I fail to see how the "progressive" model can both explain multiple extinction waves and exclude the boundary impact event as at least one of the causal factors (unless we want to go back to arguing that the crater isn't a crater and that no impact happened at the boundary in the first place). If the "progressive" model _includes_ the impact model, I have no fundamental problem with it - I just want to hear about more data from other groups supporting the claim of multiple extinction waves. "Moreover, you are saddled with the additional ad hoc hypotheses required to explain the lack of association between the event horizon and the various observed LAD's." See above. _What_ lack of association? Did you misidentify the boundary? Or do you consider no instantaneous event to ever be associable with any extinction event, just on philosophical (Popperian) grounds? "Both models are at the very least equally parsimonious, which is to say not very parsimonious at all." Again, what is the alternative model, other than the null hypothesis? On Una Smith: "How might the marine record be used to address the question of the interval of extinction of terrestrial animals?" First, if the terrestrial and marine extinctions are tightly correlated, it becomes very difficult to invoke causal mechanisms that operate only on land _or_ in the sea. Second, if the correlation holds and one record _or the other_ (here, the marine record) shows short term biotic changes, then the causal mechanism for _both_ records must be short-term. "Don't the global circulation models include several cycles between the oceans and atmosphere? What sorts of impact might anoxic oceans have on CO2 and O2 content in the atmosphere, and pH of rainfall?" My impression is that the GCM'ers are having enough trouble getting their models to work on the Eocene just by assuming a reasonably normal ocean that forces atmospheric patterns but is independent of them. Putting in realistic ocean-atmosphere feedback and anoxia as well is probably beyond our capability right now. Comments, anyone? "Can any biotic evidence help us pick out the One True Cause?... What does the extinction of X fauna (B) imply about either an impact (A) or massive volcanism (A')?" Arguments mentioned above (rudists, New Zealand plants, high-latitude forams, North American vertebrates) all take this form, and Norm and I both seem to be comfortable with this style of inference (as long as the data and interpretations are sound). As for impacts/volcanoes, I see two clear differences. First, the Yucatan crater is _in the Yucatan_. Extinctions and geological disturbances should therefore be as great or greater in the tropics and in the Caribbean than elsewhere. Second, volcanogenic extinctions should involve multiple eruption events, therefore multiple extinction events, and therefore should take place over long-time scales. If not, it would have to be shown that there is a single extinction event and that a truly gigantic eruption took place at the same time, and if this had to be argued it presumably would be easy to determine whether said eruption was better correlated to the extinction than the candidate impact. There are many other differences between the models with respect to geochemistry and related environmental sequelae, many of which should have differential impacts on faunas and floras. I leave this for others to discuss.
Partial index: