![](https://sciencetrends.com/wp-content/uploads/2017/10/Capture-1.png)
Recently many countries have committed to emissions reports to climate accords informing their greenhouse gas (GHG) annual emissions, including carbon dioxide. Forest ecosystems are huge carbon sinks and with this, they substantially contribute to intensifying the greenhouse effect. Thus, there is an increasing need for quantifying carbon in forests.
The paper proposes a statistical method for classifying natural forests into four classes of carbon stock, ranging from low to very-high stock. Such classification method employs overall climate and physical (stand) variables; mean annual temperature (MAT), and mean annual precipitation (MAP) were included as climatic variables, whereas tree dimensions and density were considered as the standard ones. The paper assessed three sets of variables: i) with all variables, ii) with all variables, except mean height, and iii) with all variables, except mean height, mean square diameter of trees, and basal area (this last reflects the degree of ground occupation of trees per unit area).
These sets of variables were tested with i) MAT and MAP together, ii) MAT only, iii) MAP only, and neither. The procedure was performed in two Brazilian forest types: Atlantic Forest (similar to evergreen tropical forests) and Savanna, which is in a drier and hotter region than Atlantic Forest. As results, the best case reached nearly 100% of correct classification. Stand variables contributed significantly to successful classification. The variable mean height exerted a greater effect in Savanna forests than those in Atlantic Forest, however, basal area and mean square diameter were the most important in both biomes.
Climate variables were most helpful when stand variables were not included in the analysis; this is an expected behavior because stand variables are directly related to tree dimensions and density, which are, in turn, directly related to carbon stored on stems, branches, and crowns of trees. The climate variables attained a maximum contribution of 9.2% for the classification, a value significant at 95% probability level. In relation to the response of forests to climate variables, forests from Atlantic Forest tended to be more sensitive to both MAT and MAP, whereas forests from Savanna had no significant climatic dependence in the classification.
As shown in other research that relates forest carbon and weather, the findings suggest that highly stocked forests are negatively correlated with MAT and positively correlated with MAP, i.e., hotter and drier regions shelter forests with the lower carbon stocks. As many other researchers have found, the study points out climate warming and drier seasons as threats to forest carbon loss in the tropical biomes.
The method could be expanded to other variables besides tree carbon, such as live or dead biomass and carbon of any tree/forest component. The validation proved that the method is efficient, reaching a high degree of correct classifications even for validation data. The method is useful in the forest management by allowing to estimate carbon stocks at local and regional scales over large areas.
This study, Carbon stock classification for tropical forests in Brazil: Understanding the effect of stand and climate variables was recently published by Hassan Camil David et al., in the journal Forest Ecology and Management.