Confused? Trust Us

 

The scientific mumbo-jumbo is so dense, it would be understandable if the average reader simply surrendered him-or herself to the presumed expertise of the assembled (albeit, biased) authors. But there are some major logical problems with the results presented in this recent paper by Pauline Kamath and co-authors. Moreover, the deployed methods are yet another instance where arcane and complex models and statistical methods, founded on numerous often unmet assumptions, are offered up as some sort of unquestionable reality.

 

Problems

 

So, to get to the problems: First of all, another researcher, Craig Miller, also estimated Ne for the Yellowstone grizzly population. His work was published in the early to mid-2000s, including a dissertation published in 2004 (for a summary of his results relative to those of Kamath et al, follow this link). He estimated that Ne for Yellowstone’s grizzly bear population was around 80 between roughly 1900 and 1980. Kamath and colleagues don’t dispute this. But Craig also estimated Ne for the 1990s using data collected from Yellowstone bears during 1992-1999. He came up with an Ne for this period that was a bit more than 100. On top of this, consider that the mark-resight estimates of population size for Yellowstone’s grizzly bears suggests that there was no increase (possibly even a decrease) between 2002 and the present (see The Trend Game). Remember that these mark-resight estimates are recommended by the researchers on the Kamath paper.

 

Bears Breeding Like Rabbits

 

Here’s the logical problem: If you accept an Ne of 100 for the 1990s (Craig Miller’s estimate) and accept that there has not been an increase in the Yellowstone population since 2002, then the 4.5-fold increase in Ne billed by the Kamath paper would have had to occur in the 7-year period between roughly 1995 and 2002. That is roughly a 21% per annum rate of increase…which is unheard of in grizzly bear populations. The highest rates of increase ever recorded have been closer to 10-15% per annum. And there is no indication, even when using biased government estimates of trend (see The Trend Game), that trend of the Yellowstone grizzly bear population has been anywhere close to this high during the advertised period. Even the most optimistic (albeit biased) estimates of growth are nearer 7%.

 

A Conundrum

 

The only way out of this conundrum for delisting advocates is to either reject Craig Miller’s estimate for the 1990s or reject the evidence for minimal or no population increase since 2002. The latter move would be in defiance of some pretty overwhelming evidence. So, if you rejected Craig’s estimate of 100 for the mid-1990s, you would be looking at sustained annual rates of increase for the period 1980 to 2002 of around 7% per annum. Again, even the most ardent apologists for population growth don’t claim this rate of increase for this long a period of time. In fact, they claim that the turn-around for the population didn’t happen until the late 1980s. And you would be left having to explain why you rejected Craig’s estimate.

 

Illogical Extrapolations

 

Another problem has to do with the tacit assumption, embedded in the Kamath paper results, that Ne is essentially perfectly correlated with N over time. Meaning, that because (presumably) Ne has increased by 4.5-fold, then total population size did as well. Which is what bear managers have quickly claimed. This gets us into The Numbers Game.

 

Pretty much all of the available population estimates suggest that the Yellowstone population grew little if at all up until the 1990s (see The Numbers Game). However, there are two population estimates from the 1990s that provide a helpful benchmark. One is from around 1992, and came in at 325 (with a huge confidence interval). Another was made for 1997, and came in around 420. So lets’ say that the population itself (N) grew by 4.5-fold since either 1992 or 1997. This would mean a current population of between 1500 and 1900 bears, and a growth rate between either 1992 or 1997 and 2002 of around 15-30%. All of these numbers are totally out of whack. Not even close to what’s biologically possible, or in line with any other available evidence. The highest population estimates currently being bandied about are 1000-1200 and (as noted earlier) the most optimistic population growth rates come in at only 4-7%.

 

The Problem of Models

 

The most logical conclusion from all of this is that there is something fundamentally flawed about the research reported in the Kamath paper. Is this possible? Answering this question gets into the nature of the methods used to estimate Ne and the assumptions upon which those methods and related models are built. Suffice it to say, the methods are amongst the most arcane out there and built on a veritable tower of assumptions about genetic, demographic, and sampling processes. For a brief overview see the paper by Hare and colleagues. Even a cursory overview of how an analysis might go wrong gets into the deep dark weeds. But perhaps it is sufficient to say that the methods and derivative results are pretty far abstracted from any tangible reality. So, yes, the analysis done by Klamath warrants skepticism on the basis of first principles and the logical violations outlined above.

 

Let's Play Monopoly

 

But this critique needs to end, yet again, with the problems that inevitably arise whenever any group of scientists holds a monopoly over scientific inquiry. In short, such circumstances debar any confidence in the results produced by such a group—such as the IGBST and the scientists it invites to participate. The only way to produce reliable scientific results is to have multiple independent (and contrary) scientists double-checking each others’ work, whether by repeating experiments or reanalyzing data in instances where there is only one possible dataset. And such is the case for Yellowstone’s grizzly bears, where there is only one Yellowstone population, and only one set of data. The extent to which the IGBST and its primary funders have fought to maintain a monopoly can only invite suspicion, especially in instances such as this one where some of the results defy logic and any other available evidence.