Carl Zimmer has a useful summary of the recent Nature paper that links global warming with frog extinctions. Brian Schmidt comments:

I was curious about the fact that none of the climate-focused bloggers on my blog roll had written on the subject. Then it occurred to me – they’re climatologists, not biologists, so they decided not to write about something outside of their expertise.

Pat Michaels wasn’t bothered by this. In fact, he reckons that he’s demolished the paper:

I have to say that this was the easiest shoot-down in some time,
because, in my humble opinion, it was the worst paper that Nature
has published on global warming. Hot rumor: investigative
journalists are on the scent that Nature and Science are skewing
their reviews. Stay tuned.

Hmmm, OK, here’s the meat of his criticism:

i-0095a303ccdbb05044711e1a46646d49-michaelspounds1.pngFigure 1 is taken from the Pounds et al. manuscript. It shows the distribution of amphibian populations as a function of daily average maximum and minimum temperatures. It also shows the original range of the chytrid fungus.

Pounds et al. cite climate data from Columbia and Venezuela showing a decline in daily maximum temperature of 0.6˚C between the 1941-70 average and the 1981-90 average and a rise in daily minimum of 1.0˚C for the same period.

We have imposed these changes upon the Pounds et al. data in Figure 1. The number of new populations possibly effected by chytrid infestations as a result of the decrease in daily maximum temperature is two and it is four for the increase in daily minimum. There are 50 populations, meaning that 12% more of the populations are subject to chytrid infection.

Unfortunately, Michaels has made some serious errors.

  • Michaels has “imposed” the changes in the temperature range on the graph by increasing the range of temperatures where the fungus thrives. This is obviously wrong.
  • Pounds’ figure does not show the distribution of amphibian populations at all. Oddly enough, the researchers did not set up a weather station at each location where frogs had been observed. The graph shows how daily minimum and maximum changes with altitude. The dots are weather stations, not amphibian populations.
  • Even if he was correct on points one and two Michaels’ conclusion about the percentages of populations affected does not follow. He assumes that the only way that a population could be affected is if the daily minimum or maximum falls into the pink band when it did not do so before. But this ignores the fact that each day the temperature varies between the minimum and the maximum. The correct question to ask, is what percentage of the time does the population spend in the pink band? This is related to the fraction of the interval from a filled circle (daily minimum) to the corresponding open circle (daily maximum). This fraction has increased for all populations below 3000 metres of elevation, so it is possible that a much higher percentage than the 12% that Michaels calculated could be affected

Of course, the errors in Michaels’ criticism have not stopped the usual folks from touting it as debunking the study: for example, Steve Milloy in this Fox News article.

In previous posts about Michaels I showed how he


  1. #1 James Annan
    January 15, 2006


    In my view one “meaty” part of his criticism was the suggestion that the fungus was recently introduced, probably by humans. I’d be interested to see any informed discussion of that point. I’m obviously not a fan of most of Michaels’ work (eg see here) but the reference he provides to back this up seems credible at first sight.

  2. #2 Tim Lambert
    January 16, 2006

    I had a quick look at the references and the strongest evidence that the fungus was recently introduced was that the extinctions only happened recently. The authors of the paper Michaels cites also suggests that perhaps climate change and fungus introduction are working together to cause the extinctions. The correlations between warm years and extinctions that Pouns et al found do suggest that climate change is a factor with fungus introduction making the frogs particularly vulnerable.

  3. #3 coturnix
    January 16, 2006

    So, were the fungi always there, but warmer climate makes it more numerous/aggressive, or did the fungus recently migrate into the frog territory following its own temperature cline?

    [Posted from the dreadful IE, as I cannot comment from Firefox]

  4. #4 James Annan
    January 16, 2006

    On further investigation, it seems to be widely accepted that the fungus was introduced recently, despite Tim’s one-eyed presentation of the evidence. But I don’t see any mention of this in the paper itself, let alone the discussion (eg Carl Zimmer) that has followed it.

    “fungus introduction making the frogs particularly vulnerable”

    ROFL! It’s this “virulent pathogen” that kills them, and the disease only started spreading worldwide after international trade in its original host (which it doesn’t seem to affect much at all) started up last century.

    As you know, I’m no fan of much of Michaels’ output. But the public presentation of these results is simply unacceptable in my opinion.

  5. #5 Tim Lambert
    January 16, 2006

    I don’t think it is a good defence of Michaels to bring in papers that he didn’t cite. From one he did cite (Daszak 2003)

    The data suggest that Batrachochytrium 1) may be endemic to these regions and the amphibian deaths and declines attributed to it have only recently been discovered; 2) may be endemic and has recently become pathogenic (e.g., through an increase in the organism’s prevalence or virulence, or a decrease in the host’s defenses), or 3) may have been introduced recently into these geographic regions and is now parasitizing novel host species.

    James, why don’t you write up a criticism? I think you could do a much better job than Michaels.

  6. #6 James Annan
    January 16, 2006


    I’m not so much trying to defend Michaels in particular as simply work out what is going on. I wonder if the human-mediated spread of the disease is such a well-established fact that the authors didn’t even think it worth discussing. Anyway, if true, it certainly puts a very different complexion on things IMO. I may write something about it the next day or two.

  7. #7 Tim Lambert
    January 16, 2006

    They write “Although pathogens are implicated[14-28]” and the papers arguing for human-mediated spread are included in those 15 references. So maybe they expect you to be familiar with the various theories.

  8. #8 Eli Rabett
    January 16, 2006

    Michaels’ argument about cloudiness was interesting. I started looking at the ICCP database but am not expert enough in it to go much beyond the map he showed. I am distrustful enough of Michaels based on past performance not to take him at face value

    Plaese bling bact the spel ckr.

  9. #9 Chip Knappenberger
    January 16, 2006

    Mr. Lambert,

    Thanks for your careful read of our posting on concerning our critique of the Pounds et al. Nature article on amphibian extinctions. You are correct in that we made a mistake in interpreting the points on Pounds et al. Figure 4c. We have subsequently revised our original posting. The revised version is available at
    But, please realize, that the major point we are trying to make does not hinge on this one detail.

    And also thanks for the thoughtful comments from James Annan and Eli Rabett for actually taking the time and effort to look into our cited sources rather than simply dismissing our claims based soley upon who the delivery person was. The Pounds et al. study postulated a complex solution to a problem with a seemingly simple answer–a recently introduced pathogen. It was a bit disconcerting to see such a stretch (very little climate data, very sparce amphibian data) make it to the pages of Nature. That was our main point.

  10. #10 Eli Rabett
    January 16, 2006

    Since it is afternoon I will engage in a bit of snark: There is no problem so complicated that it does not have a simple but wrong answer.

  11. #11 Steve Bloom
    January 17, 2006

    While we’re being snarky, it’s probably worth pointing out that Chip and Pat’s gratitude did not extend so far as to give proper credit to Tim.

  12. #12 Jack
    January 17, 2006

    The ISCCP cloud data resolution troubles me a great deal. I found this reference:

    High-Resolution Daytime Cloud Observations for Northwestern Mexico from GOES-7 Satellite Observations

    which contains the following final sentence in the abstract:
    “The derived high-resolution cloud estimates, when compared with similar estimates from the International Satellite Cloud Climatology Project (ISCCP D1), were about half those obtained with the low-resolution data, indicating that, in this complex study area where land and water boundaries are in close proximity, low-resolution satellite observations of clouds may not be able to depict the true cloud cover.”

    For an area like Costa Rica, where “land and water boundaries are in close proximity” (one would think) — and I would add, a wide range of hydrologic conditions from the coastal lowlands to the Cordillera peaks and cloud forests — I don’t think that the ISCCP resolution is sufficient to draw conclusions like those in the WCR piece. The kicker is: the ISCCP D1 spatial resolution is 280 km (quoting from the paper). Does Costa Rica even fill a single ISCCP grid cell? And Michaels et al. are trying to draw conclusions about morning fog in a cloud forest that is a very small part of the whole Costa Rican cordillera? Hmpfh.

    Par for the WCR course, unfortunately.

  13. #13 Jack
    January 17, 2006

    On further review… Costa Rica is 51,100 square km in area. The paper says that the ISCCP D1 resolution is 280 km. This means that the side of a grid cell is 280 km, i.e., that the grid cell is 280 x 280 km, or (drum roll) — 78,400 square km. [Figure 11 in the paper is wonderfully illustrative.] So Costa Rica isn’t big enough to entirely fill a single ISSCP D1 grid cell. Any inferences based on that data may therefore be inaccurate (particularly if the source wants to be misleading).

  14. #14 Chip Knappenberger
    January 17, 2006


    At least give us some credit. We made our comments based upon actual cloud data–whether they are of the appropriate scale or not is a legitimate concern–but at least we provide some data! (also note that from our trends map there are really no appreciable trends anywhere in the overall Costa Rican region–but, obviously things may be different at a sub-resolution level). Pounds et al. provide NO CLOUD DATA, and, in fact, in their 1999 paper, they suggest a rising cloud level is killing off the frogs–a rising cloud level would not lead to warmer nights and cooler days–a somewhat opposite conclusion than they are pitching this time around. So while our data may be less than ideal, Pounds’ data is totally absent.

  15. #15 Jack
    January 17, 2006

    I give you credit for providing data on cloud cover. I don’t give you credit for omitting to mention that the cloud cover data you provided was probably of minimal value for evaluating the relationship described in the current Pounds paper. Your critical point makes the ISCCP D1 data seem relevant — it’s probably not, and Pounds et al. might therefore be credited for not including irrelevant data in their manuscript.

    However, the difference in cloud mechanisms described in the 1999 paper and this paper — as you mention — is worthy of examination. Because the cause of amphibian loss proposed in 1999 is different that the one proposed now, it would be truly meritorious if someone would actually try to determine how warming affects cloud cover in high-altitude rain forests.

    I did a rapid search and determined that the likeliest trend in these mountain rain/cloud forests is less cloud cover (and higher condensation level), based on Lawton, Nair, Pielke, and Welch in 10/21/2001 Science (as quoted by National Geographic, but the actual paper is linked below) —

    — a trend exacerbated by lowland deforestation, so it is not strictly a temperature-related trend. Lawton et al. 2001 is in accord with the Pounds (1999) paper, and not the current paper. Sooo… while the observed temperature trends seem clear enough, their cause is murkier. I therefore wonder if this lack of clarity is related to the lack of certainty regarding the “global dimming” issue and its relationship to GHG-induced climate change.

    Perhaps the loss of mist in the cloud forests is still a viable explanation for the amphibian losses. And I will speculate here — maybe moisture-stressed amphibians are more prone to fungal infestations. ???

New comments have been temporarily disabled. Please check back soon.