Figure 5
Schematic illustration of biases in reconstructions of black spruce growth. All trees estimated to have been once present in the population have been considered, including in green, trees sampled in this study (same as in Fig. 3) and in red and blue respectively, trees that were excluded because they were smaller than 91 mm at the time of sampling (cf. “big-tree selection bias”), and because they died before sampling (cf. “slow-grower survivorship bias). The black line shows the unbiased growth reconstruction of all trees historically present in the simulation (all dots), the red line shows the trend arising due to the slow-grower survivorship bias, and the blue line shows the bias due to the big-tree selection bias. This analysis shows that the spurious trends from Fig. 3 are for about 80% driven by the big-tree selection bias and for the remainder by the slow-grower survivorship bias. A subset of only 1 million randomly selected individual values are shown for clarity.
