Quantifying forest change.

In a recent issue of PNAS, Hansen et al. (1) did an excellent job of arguing for the need for a more consistent data set to investigate changes in global forest cover. The indicator they chose, gross forest cover loss (GFCL), marks an improvement both in reproducibility and comparability. However, it does so by using data that the authors themselves state “captures only part of the global forest cover change dynamic” (p 8651). One would expect that an analogous measure of forest growth over the same period might show a gain in forest cover if data on loss were excluded. In describing the forest cover dynamic, Alig (2) comments that “Net changes (area into forest minus area out of forest) are typically much smaller than total or gross changes (area into forest plus area out of forest).” The proposed indicator, the GFCL, differs markedly from other forest coverage indicators used at the national and international level. The reader is led to believe in the superiority of this data set as it directly correlates with the “biophysical presence” (p 8652) of trees, implying that other definitions are simply random or wholly bureaucratic in nature. Thus, national and international sources that lead to conclusions other than those of Hansen et al. (1) are brought into question. The measure that Hansen et al. (1) used to define forest cover does not consider reforestation processes or provide a proxy for carbon uptake. In fact, many young forests that have not reached the threshold of 5-m height of Hansen et al. (1) are growing rapidly and thus are absorbing carbon rapidly. Volumetric data on cubic feet of forest inventory also tell a different story. For example, in the state of Louisiana, an area marked as having high forest cover loss by Hansen et al. (1), the US Forest Service reports that, between 1991 and 2005, growing stock volume increased 8.9% (3). It is tree biomass, heavily represented by growing stock volume, of course, that sequesters carbon from the atmosphere. The results put forward by Hansen et al. (1) are based on a 5-y time span. Given that the period for forest coverage cycles spans decades, changes over a single 5-y span are simply not representative of secular trends extending beyond a part of the longer natural cycle. No single ideal measure can convey the health of the forest systems. The use of consistent data sets is welcome; however, caution is required in deriving indicators from them. The indicator that Hansen et al. (1) used would seem to exaggerate global forest loss.