What was done
The authors used a process-based ecosystem model that makes monthly calculations of carbon and nitrogen fluxes and the pool sizes of various terrestrial ecosystems to compute global net ecosystem production (NEP) from 1990 to 2100 based on three different scenarios of atmospheric CO2 and temperature change.

What was learned
For the scenario with lower-than-typically-assumed increases in CO2 emissions and air temperature, the NEP of the globe increased from a positive (carbon-sequestering) value of 0.8 Pg C yr^-1 in 1990 to 1.3 Pg C yr^-1 in 2100, for a rate increase of 62% over the 110-year period.  For typically-assumed increases in CO2 and temperature, corresponding NEP values were 0.8 and 2.6 Pg C yr^-1, yielding a rate increase of 225%.  And for higher-than-typically-assumed increases in CO2 and temperature, NEP values of 0.8 and 3.4 Pg C yr^-1 were obtained, for a rate increase of 325%.

What it means
Based on real-world knowledge of how earth's plant life responds to increases in air temperature and atmospheric CO2 concentration (or the best approximations of it available at the time), this study predicts dramatic increases in biospheric prowess, as recent historical trends of these two parameters continue into the future.  Hence, it is not surprising that the authors state that "caution should be taken in using these NEP results in policy discussions relevant to anthropogenic CO2 emissions and carbon taxes."  Sure.  If they were included, it would indicate that carbon taxes to reduce CO2 emissions are totally absurd, as CO2 emissions - and even temperature increases - are a tremendous help to the biosphere.  And that admission would not be politically correct.