What was done
The authors used an ecosystem model to investigate the impacts of observed changes in growing season length over an 88-year record (1900-1987) for twelve sites in the eastern deciduous broadleaf forest of the United States.

What was learned
For individual sites, the length of the growing season regularly varied by more than fifteen days; and for each one-day increase in growing season length, net ecosystem production rose by 1.6%, while evapotranspiration rose by only 0.2%.

What it means
Recent observations of seven- to eight-day increases in growing season length in high northern latitudes over the past decade or so have been suggested to be the result of CO2-induced global warming.  However, in the words of the authors, these increases are "neither unusual nor necessarily a sign of permanent climate change."  Indeed, they note that ten-day growing season length decreases were characteristically observed over periods of one to two decades throughout their 88-year study period, while increases of the same magnitude occurred in as little as four to six years.  The authors additionally conclude that "simulations attempting to investigate future long-term climate warming effects on the carbon cycle should clearly use a dynamic, climatically controlled growing season length" (instead of the static one that is currently used), because of the great impact that a lengthening of the growing season has on the sequestration of carbon.  As they truly note, "persistent increases in growing season length may lead to long-term increases in carbon storage," which, of course, tends to counterbalance - and, indeed, overpower - the effects of increasing air temperature on respiration rates that return CO2 to the atmosphere.  Once again, therefore, we have another example of a phenomenon that counters the threat of a runaway greenhouse effect.