Do standard tropical forest monitoring protocols bias height–diameter allometries?

Abstract

The widely used protocols for monitoring tropical forests exclude trees with certain characteristics when measuring tree heights. Here, we estimated the prevalence of exclusion characteristics (Broken, Fallen, Leaning and Rotten) and the effect on tree height estimates for trees in a 1-ha plot in the mountainous Atlantic Forest. We recorded the prevalence of exclusion characteristics from the 801 trees in the plot and then fitted an allometric Height–Diameter at Breast Height (i.e., at 1.30 m; DBH) (H:D) model from a subset of 81 trees to establish if the exclusion characteristics had an impact on the allometric H:D relationship. Of these 81 trees, 40 had no exclusion characteristics (i.e., they were the “standard” trees according to the protocol), and 41 presented at least one of the exclusion characteristics (i.e., “not standard” trees). We found that the exclusion characteristics were exhibited in 252 (31.5%) of the 801 trees. Leaning was the predominant characteristic (28.3% of the 801 trees), while the others were recorded relatively rarely (below 2.5% per characteristic). For both the standard and non standard trees, tree height increases systematically with increasing DBH. However, the H:D relationship was significantly different between the two sets of trees: For given DBH, the height of non-standard trees was significantly lower than the height of standard trees. Given the high prevalence of exclusion characteristics, and that there is a significant difference in the H:D relationship, our results indicate that protocols selecting trees of a certain standard for height measurements significantly overestimate the tree height and thereby estimates of biomass and carbon stocks.