Extended Data Fig. 2: Effect of plot area, aggregation procedure and plot design on estimates of AGC stocks across the AfriMont plot network.
From: High aboveground carbon stock of African tropical montane forests
a, Relationship between AGC stocks and plot area of plots before aggregation. The red line shows the fit of a locally weighted regression model (span 0.75) relating these variables, with dashed lines showing the standard errors. b, Variation in AGC stocks using either all plots before aggregation (unaggregated), plots before aggregation but excluding those <0.2 ha (unaggregated, >0.2 ha) or the aggregated plots used in the main analyses (aggregated). c, Effects of plot design on AGC stocks (each site represents one dot). Sampling strategies include random or stratified random, plots positioned along transects, plots established within elevation bands, subjective measures such as choosing an area of forest considered representative of the wider area, and other strategies (one plot sampled per site or unclear strategy). Carbon stocks (log transformed) did not differ significantly between sites with different sampling strategies (analysis of variance F4,39 = 0.432, P = 0.785). For specific site information, see Supplementary Table 5.
