By Edward Gregr and Kai Chan
Too often, the fruits of years of
research effort yield peer-reviewed publications that are greeted by the
resounding echo of … silence.
We were therefore extremely
pleased to find our recent article was not only read, but genuinely understood
and actively supported by the most exacting of all readers, the Editor in Chief
of the highly respected journal BioScience. The article in question was our
recent review article entitled Leaps of
Faith: How implicit assumptions compromise the utility of ecosystem models,
wherein we examined how assumptions and uncertainty were treated by the 60 most
popular articles describing ecosystem models that were intended to inform
management.
In his editorial Models are Not Toys, Dr. Timothy
Beardsley focused squarely on our article, (re)articulating our main points
with such eloquence that it was clear he fully and completely "got
it". Rather than confuse his message by paraphrasing, we simply reproduce
the bulk of that editorial directly.
"Mathematical modeling has
become central to much of biology and related sciences, to the extent that
ecosystem studies that omit to describe a model of their phenomena of interest
might even seem underdeveloped. The power of well-constructed models hardly
needs defending—their ability to provide a vivid sense of what is going on in a
complex situation is felt intuitively. Yet the fascination of modeling can
easily lead the unwary into a fantasy of neatly interacting parameters that
cease to meaningfully represent the messy, hard-to-measure real word. The
temptation is to test just the most interesting model predictions and
assumptions—the ones that yield the results we hope to find—and to forget
possibly unwelcome complications at the interpretation stage.
For sure, there may be good
reasons to ... develop a model that captures some important features of the
world even when the model is incomplete in significant ways. Science has made
vast strides by employing this reductionist strategy, as Gregr and Chan
acknowledge in their important article ... Yet when attempts are made to draw
management conclusions from models that are described without a full account of
the assumptions and uncertainties that went into them, danger lurks.
Gregr and Chan's survey of
articles purportedly relevant to ocean ecosystem management yielded results
that can fairly be called “shocking.” “Ecosystem” articles left all assumptions
implicit 39 percent of the time, and “social–economic” articles articulated no
assumptions 68 percent of the time. The problems may be linked to the
increasing fragmentation of modelers into specialized communities that are
familiar with their own assumptions—but fail to explain them to outsiders.
Gregr and Chan also remark on the “widespread faith in data quality and
suitability.” Understandable, given the difficulty of gathering custom
environmental data, but faith should have no place in science.
The authors provide a valuable
typology of assumptions and uncertainties that all ecosystem modelers would do
well to use as a checklist when planning and reporting their research. Unstated
assumptions and uncertainties, besides causing unwise management actions, could
bring this vital field into disrepute. And there is no reason to think the
ocean scientists are more at fault than those who study other ecosystems."
We could not have said it better
ourselves. But we are thrilled we articulated it clearly enough, and that this
topic was seen as critical enough, for at least one senior Editor to find the
work compelling enough to recognize the importance and contribution of our
review. It is that sort of validation that keeps scientists motivated, and
makes us feel that our work, however slowly, is making its way into the fabric
of scientific understanding.
The complete editorial is
accessible online
at BioScience. The article is available on request from EG.
Beardsley, T. M. (2015). "Models Are Not
Toys." BioScience 65(1): 3.
Gregr, E.
J. and K. M. A. Chan (2015). "Leaps of faith: How implicit
assumptions compromise the utility of ecosystem models for decision-making."
BioScience 65(1):
43-54.
EG/KC