I’ve wasted a lot of time recently puzzling over some of the latest federal waterfowl statistics.

Take these, for example: In 2007, California led the nation with a total harvest of 1,632,896 waterfowl.  Mallards constituted 270,300 of that total. Louisiana came in second with a kill of 1,532,800, including 142,526 mallards. Arkansas claimed third place with a total of 1,112,200, including 578,131 mallards. On the Pacific Flyway, Oregon hunters took 684,198, with 271,041 mallards; and Washingtonians bagged 452,598 total birds of which 235,351 were mallards. Or so they say.

Does anything strike you as odd about this collection of numbers? Am I alone in wondering how the authorities can declare the kill down to the last duck in most instances – of which these are but a minor sampling of represented exactitude?
 
Of one thing I am sure. No one asked me – this year or ever -- for a definitive report of my personal kill. Frankly, that oversight fills me with resentment. The feds obviously have to be getting highly exact numbers from canvassing somebody. How did I get left out?

Reminds me of the feelings I had in high school when my buddy, Meatball Mitchell, got a regular supply of comic books from a really cool grandma, a woman who indulged in an occasional chaw of Redman chewing tobacco when no one was looking, had run a speakeasy during Prohibition and understood the literary needs of young boys.  My grandparents had not run any speakeasies. They believed that comic books rotted the mind.  They had brainwashed my parents into acceptance of such arbitrary silliness. I struggled under the weight of that unjust deprivation, the sense of being shortchanged in the cool grandma department and ignored in a matter of importance. And now, I suffer a repetition.

The U.S. Fish and Wildlife Service doesn’t want my numbers, even as it represents to the world that it can compute the harvest in total and within a single species down to the last duck. As I always end up with quite a few, those numbers have to be short by that many -- and by those of the other neglected hunters of my acquaintance as well. My inquiries (unscientific and anecdotal, of course) suggest that nobody else got asked for their exact counts either.

I disregard that small questionnaire we answer when we get our licenses. You know that one. You must estimate your kill during the season that ended at least eight months earlier. The choices are: 0 to 10, 10 to 20, 20 to 30, and over 30. No differentiating by species and no allowance for imperfect memory, a process that treats the hunter who kills 31 equally with the hunter who kills 150.  I certainly hope the feds are not constructing their claim of exactitude on that sort of “scientific information.”

It is worthwhile to consider how one would construct a model that might lead to such a compendium of precision. Start with what you would need to know: (A) How many birds begin the fall flight? (B) How many of those do hunters kill? (C) How many die of natural causes? The breeding population the following spring should equal (A) minus (B) and (C).  We are told the field counts of breeders constitute the most accurate data biologists produce (subject to an acknowledged margin of error), but I have never seen a meaningful, straight-forward attempt to reconcile that number with fall flight, less kill, less natural death. To close the circle, add to the breeding count the ducklings produced, deduct a summer attrition factor – and bingo, the next year’s fall flight number emerges.

Look carefully and ask yourself this question: Do we have any inputs sufficiently accurate to construct and use such a model as a management tool to assure preservation of a critical mass of breeding birds?  We do not. But the most arresting fact is this: when you apply the statistics the regulators’ actually use within the framework of a straight-forward population model, they don’t compute. Not even close.

Let’s illustrate with numbers that pertain to California mallards, a conveniently finite population that predominantly breeds locally.
 
The official count proclaims that hunters killed 270,300 mallards during the season that began in October of 2007. If you allow for what is described as scientifically determined natural attrition during the fall and winter, this means that our California mallard fall flight in 2007 had to contain 1,162,060 mallards. Not just ducks. Mallards.  The survivors – the breeding bird population in the spring of 2008 – equaled 297,100.
 
Nothing in the literature suggests that we have ever had that many since anyone starting keeping official estimates.

You get the same discordant result if you chase down any of the other numbers the authorities feed us as gospel.  For example, using U.S. Fish and Wildlife Service numbers, the leadership of the California Waterfowl Association likes to intone that we do our mallard flight no long-term harm so long as the kill does not exceed eight percent of the adult hens. That number apparently finds its genesis in band data estimates based on mid-continent mallards. Wholly apart from its geographical irrelevance for the local California mallard population, it lacks foundation on so many grounds it is hard to know where to begin.

First, at some low level of population we cannot accept any hunter kill at all if we hope to sustain a critical mass of breeders. This is not a matter of percentages but of absolute numbers of ducks.¹   If that is not the case, why did we shut down the canvasback season for 2008-09, based on the low total numbers in the breeding population? Doesn’t the same logic apply to every species and every discrete sub-population within each species, including our local mallards?

And doesn’t the very notion of harvest expressed as a percentage depend upon our having a consistently solid estimate of the total number in that segment of the population to which the percentage applies? Doesn’t the very notion of eight percent presume that we know how many adult mallard hens we have in the fall flight?
 
The service (and derivatively CWA) thinks it knows these answers based on band returns and the volunteer collections of wings, to provide both the absolute numbers and the aging data. But we have no way to know that the banded birds reflect a true sampling of the flight as a whole – and the returns, whether of band information or wings, tell us only about the birds recovered almost exclusively by hunters.  
 
Given the way they are collected, aren’t those wings just dreaded and derided “anecdotal evidence” in another form?  The wings are real enough, of course. But the collection process creates a sample with no claim to “science” whatever – no formulation other than creative surmise to connect the wings with an accurate, overall population model.

And if we seek to protect our breeding population at some minimal level of critical mass, we have that little problem of making a reasonable assessment of death by natural causes, a factor that cannot be proved by comprehensive field survey through any technique currently available. Accountants have an unflattering term for a calculation of that sort. They call it a “plug number” -- estimate the fall flight, subtract hunter kill based on the indicators we rely upon with all their glaring imperfections, subtract the spring counts and voila, you have the number that must have died by natural causes – because we have no other explanation for the discrepancy (other than the unspeakable, i.e., we, the scientists, do not know). 
 
OK.  No need to prolong this. So we have imperfect information, a compendium of estimate build on estimate, compounded by unprovable and untested assumptions. And that may be the best we can do. What’s the problem?

I see at least three, none of them concerning the unavoidable holes in our knowledge but (a) in our institutional unwillingness to acknowledge those imperfections, (b) the failure to take account of worst case possibilities in light of those imperfections, and (c) the habit of attacking everyone who dares to question the so-called “science” behind the compendium of guesswork on which we rely. And perhaps a fourth as well: the deceptive way in which the information is presented to make it appear more precise and factual than it is – as in presenting kill and other numbers down to the last duck.

In his annual statement to the shareholders of his holding company, Berkshire Hathaway, Warren Buffett made a few pungent observations about the genesis of the current economic meltdown that seem to apply here. He observed that the business and financial community fatefully ignored the perils of relying on economic models devised without worst case scenarios while relying on outdated assumptions as part of the analysis.  Americans, he wrote, have become mesmerized by “a nerdy-sounding priesthood, using esoteric terms such as beta, gamma, sigma and the like” as cover for pretending to more knowledge than they in fact possess.  His advice: “Beware of geeks bearing formulas.”  I would humbly add: particularly if those geeks construct their formulas to serve a pre-existing conclusion, and defend them against all questions in the name of science, when true science evolves and advances in response to never-ending analysis, test and questioning.

All of this takes me back to Meatball Mitchell’s speakeasy running, tobacco chewing, comic-book smugglin’ grandma. I doubt that she was a duck hunter – but she had a hard-eyed sense of reality, in defiance of convention. Faced with the situation we face in our waterfowl world, I have no doubt that she and the Sage of Omaha would both agree:  Beware indeed of geeks bearing formulas, particular those who act like a priesthood of dogma, erecting a façade of a false precision offered to mask an uncertain reality.

Finally – and just for fun – when you have nothing better to do some evening, see if you can make any of the numbers our regulators feed us come together in an internally consistent fashion within the framework of a logical model.  You can find the numbers at Madduck’s Reference Desk. If you succeed, let me know. And if you cannot make it work, let me know that also. Then there will be at least two of us who would be better off doing something else in the name of getting a life, as our children pompously intone.
___________________

¹ In a population of 200, ten dead ducks represents just five percent.  No one would claim that 190 is a safe and adequate breeding population.