Research Article

Riding the Wave: Reconciling the Roles of Disease and Climate Change in Amphibian Declines

  • Karen R Lips mail,

    a To whom correspondence should be addressed. E-mail:

    Affiliation: Department of Zoology, Southern Illinois University, Carbondale, Illinois, United States of America

  • Jay Diffendorfer,

    Affiliation: Illinois Natural History Survey, Champaign, Illinois, United States of America

  • Joseph R Mendelson III,

    Affiliation: Zoo Atlanta, Atlanta, Georgia, United States of America

  • Michael W Sears

    Affiliation: Department of Zoology, Southern Illinois University, Carbondale, Illinois, United States of America

  • Published: March 25, 2008
  • DOI: 10.1371/journal.pbio.0060072

Reader Comments (5)

Post a new comment on this article

Amphibian extinctions: Disease not the whole story

Posted by plosbiology on 07 May 2009 at 22:23 GMT

Author: Camille Parmesan
Position: Associate Professor
Institution: Integrative Biology, University of Texas at Austin
Additional Authors: Michael C Singer
Submitted Date: March 26, 2008
Published Date: March 28, 2008
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

Amphibians have been mysteriously going extinct worldwide. Lips et al.1 claim to have bolstered evidence for disease as the culprit, and refuted evidence of a climate change link. We argue that Lips et al.’s analyses of invasive spread of the fungus Batrachochrytium dendrobatidis (Bd) are questionable and that their critique of climate explanations is seriously flawed.

In attempts to discredit climate change as a causal factor in Central American extinctions, Lips et al. reveal serious statistical ignorance. Pounds et al.2 found highly significant associations between the last year a species was observed (LYO) and high temperatures in the prior year. Lips et al. criticize use of LYO as a surrogate for extinction: amphibian censuses are sporadic and sampling intensity variable, resulting in noisy datasets. However, Lips et al.’s re-analyses, which attempted to simulate uncertainty by adding large levels of error to LYO data, is statistically inappropriate. Increasing data noise (adding additional error) will eventually cause loss of any statistical relationship in any dataset. Therefore, Lips et al.'s re-analysis shows nothing beyond a known statistical property of all datasets: as random noise increases, the ability to detect a true relationship decreases.

Error in LYO should make finding a relationship more difficult, but should not bias the outcome towards a particular conclusion. Thus, the fact that Pounds et al. found significant relationships between climate and LYO indicates that the relationship was so strong that it was robust to any errors in estimates of year of extinction. Might this conclusion have been affected by bias? A systematic trend for researchers to alter their sampling in response to the previous year’s weather would do the trick. However, we find this scenario- of temperature-dependent variation in censusing techniques/intensity with 1-year lag time- to be implausible.

In advancing their argument that climate-independent spread of Bd is the cause of extinctions, Lips et al. draw maps of dates of first observed population decline (DOD), interpreting these as times of first arrival of Bd. However, DOD is not an appropriate surrogate for Bd arrival because Bd occurs in populations that are not declining3 and amphibians have disappeared in the absence of Bd4. Lips et al. postulate that Bd spread in “waves from 4 oldest records of decline.” Within 4 of 5 waves, they find a significant positive regression between distance from the oldest declines and subsequent DODs, a pattern which they argued fits a spread of disease from each focal point. But each ‘wave’ was arbitrarily delineated by starting from the oldest DOD records. Observations inconsistent with a single ‘wave’ were reconciled by adding waves until a consistent pattern was achieved. While this may correctly represent events, this procedure weakens the impact of statistical significances.

In conclusion, the Pounds et al. study2 shows that patterns of amphibian extinctions (LYO) in space (elevation) are moderately consistent with climate change being the driver, and that amphibian extinctions (LYO) in time are strongly consistent with climate change being the driver (extinctions being high after hot or dry years)2,3. The Lips et al. study1 shows that spatiotemporal patterns of amphibian decline (DOD) are moderately consistent with disease spread (Bd) being the driver. These two hypotheses are both supported by multiple studies, are not mutually exclusive, and may in fact be interactive.

1. Lips KR, Diffendorfer J, Mendelson JR III, Sears MW (2008) PLoS Biol 6(3):e72. doi:10.1371/journal.pbio.0060072

2. Pounds JA, Bustamante MR, Coloma LA, Consuegra JA, Fogden MPL, et al. (2006) Nature 39: 161–167.

3. Retallick RWR, McCallum H, Speare R (2004) PLOS Biol 2:e351. doi:10.1371/journal.pbio.0020351

4. Pounds JA, Fogden MPL, Campbell JH (1999) Nature 398: 611–615.

No competing interests declared.