Blog‎ > ‎

Using Uncertainty to Improve Natural Resource Management

posted Mar 31, 2012, 10:42 AM by   [ updated Feb 7, 2013, 8:25 AM ]
Uncertainty often feels like a problem to be solved:  if we can reduce uncertainty, we can improve the decision-making process.  Natural resource managers confront uncertainty on a regular basis.  They deal with large complex systems - entire landscapes - that are dynamic and often poorly understood.  

Simultaneously they must continually monitor risk:  Risk to the products, services, or populations they are tasked with managing.  In the western US, wildfire poses perhaps the largest immediate risk to their management objectives, but there are many others including climate change, invasive species, and so on.

The true risks, however, are often obscured by uncertainty.  In one of my favorite papers, Higgins et al. (2003) broke uncertainty into three components:

1.  Model uncertainty.  Do we have a good understanding of the world?  Are we able to logically or mathematically represent the world in order to use what we know (data) to estimate the unknown?

2.  Parameter uncertainty:  Do we have enough good data?  This is where natural resource managers are particularly challenged.  Much of what they’re tasked with protecting is rare or reclusive.  No where is this more true than when managing populations of endangered or threatened or rare species.  Often little is known about how many there are, how fast they breed, how they die.

3. Inherent uncertainty.  The world is a stochastic, unpredictable place.  Nassim Nicholas Taleb (The Black Swan) provides a brilliant example using pool:  One would need to know the gravitation pull of the pool players themselves to be able to predict beyond the 8th bounce or hit during the break.  Now imagine predicting where lightning will strike a forest next year or 10 years from now.

Science can reduce the first two sources of uncertainty through diligent hypothesis testing and better monitoring.  We are stuck with the third.  The inherent uncertainty of a system can never be reduced.  Some systems have more, some less.  And large inherent uncertainty reduces the aggregate effectiveness of management:  their actions may be nullified by a rare, unpredictable event, such as an intense crown fire.

How does this help or inform managers?  My conjecture is that scientists can help natural resource managers optimize uncertainty and risk management.  

My conjecture is that uncertainty follows a non-linear relationship with risk (see figure).  There is often large uncertainty around small events that pose little risk to overall management objectives.  I call these ‘flock of chickadee’ events.  A perfect example is the mortality of individual trees (aside from major disturbance events).   Individual trees die for a multitude of reasons that are often poorly understood and are generally unpredictable.  

At the other extreme are the Black Swan events:  extremely rare, unpredictable, and very harmful to management objectives.  Large crown fires are a classic example.  Tornados, hurricanes, and earthquakes fall into this category.

The question and challenge:
 Can we help managers locate the risk/uncertainty sweet-spot?  This is where uncertainty is manageable (not too high) and the risk is high enough to justify taking action - active management.  Locating the sweet-spot will help managers optimize their allocation of scarce (and declining) resources and will require identifying the proper scale(s) or location(s) for effective management.  As scientists, I don’t believe we’re there yet.  But the models and data are rapidly becoming available to get us closer to this goal.

- Robert Scheller

Higgins SI, Clark JS, Nathan R, Hovestadt T, Schurr F, Fragoso JMV, Aguiar MR (2003) Forecasting plant migration rates: managing uncertainty for risk assessment. J Ecol 91:341–347