Does the disturbance hypothesis explain the biomass increase in basin‐wide Amazon forest plot data?

Abstract

Positive aboveground biomass trends have been reported from old-growth forests across the Amazon basin and hypothesized to reflect a large-scale response to exterior forcing. The result could, however, be an artefact due to a sampling bias induced by the nature of forest growth dynamics. Here, we characterize statistically the disturbance process in Amazon old-growth forests as recorded in 135 forest plots of the RAINFOR network up to 2006, and other independent research programmes, and explore the consequences of sampling artefacts using a data-based stochastic simulator. Over the observed range of annual aboveground biomass losses, standard statistical tests show that the distribution of biomass losses through mortality follow an exponential or near-identical Weibull probability distribution and not a power law as assumed by others. The simulator was parameterized using both an exponential disturbance probability distribution as well as a mixed exponential-power law distribution to account for potential large-scale blow- down events. In both cases, sampling biases turn out to be too small to explain the gains detected by the extended RAINFOR plot network. This result lends further support to the notion that currently observed biomass gains for intact forests across the Amazon are actually occurring over large scales at the current time, presumably as a response to climate change.

Land-use and climate change risks in the Amazon and the need of a novel sustainable development paradigm

“…Terrestrial ecosystems and, in particular, tropical forests have been recognized as strong carbon sinks, comparable to the oceans (132)(133)(134). However, the potential connections between sinks of carbon and the spatial patterns of soil and tree characteristics across the region are also noteworthy.…”

Detecting and projecting changes in forest biomass from plot data

“…Fisher et al (2008) also use a flawed procedure to fit the size distribution of disturbance events, and thus almost certainly overestimate the probability of large disturbance events. Gloor et al (2009) corrected these errors, and concluded that sample sizes in area and time were sufficient to demonstrate that average biomass change in 135 forest plots, typically 1 ha or larger, across the Amazon basin is significantly positive, under the assumptions that plots were an unbiased sample of the landscape and that individual plot biomass change estimates were unbiased. It is important to note that these results are sensitive to the frequency of large disturbance events, and that there are very limited existing data relevant to estimating these frequencies.…”

“…Fisher et al (2008) and Gloor et al (2009) recently explored the influence of sampling error on true power to reject the null hypothesis of no change in biomass given the potential for large-scale disturbances. These studies use simulation analyses to explore the importance of clustered disturbances combined with the 'slow in, rapid out' nature of biomass change, and the sample size needed to confidently reject a null hypothesis of no change.…”

“…Mascaro et al 2011a), and thus LiDAR is well placed to provide information on the spatial autocorrelation structure of biomass at landscape scales. Both LiDAR and other remote-sensing tools can provide landscape-scale data on canopy disturbance rates and the size distribution of disturbance events -critical quantities for determining the true power of a given plot sampling effort (Fisher et al 2008;Gloor et al 2009). However, most remote-sensing tools are likely to have limited if any ability to detect small increases in biomass in highbiomass forests (because most remotely sensed metrics saturate at high biomass), limiting their usefulness for quantifying potential subtle changes in oldgrowth forests.…”

“…Thus, detecting global change in old-growth forests requires a large enough sample size with adequate distribution in space and time to be able to confidently distinguish global change influences from the natural gap-dynamic cycle and cyclical climate variation. The question of what constitutes an adequate sample size has been addressed in the recent literature (Fisher et al 2008;Gloor et al 2009). …”