Quantifying the effects of topography, substrate, and land-use history on aboveground biomass in a California ecosystem

Background/Question/Methods   Measuring accumulated biomass is not only important for carbon accounting; it offers an integrated measure of many different ecosystem processes. However, at the landscape scale, controls on biomass accumulation are poorly understood. In this study we asked the questions: How well can aboveground biomass (AGB) be predicted using airborne remote sensing? and How much of the variation in AGB can be explained using readily available environmental and land use data sets in a mature ecosystem? To address these questions we chose Jasper Ridge Biological Preserve (Woodside, CA) a study site, as historic aerial photographs indicated that little change has taken place at the site since at least 1928. We then combined field measurements and high resolution data from the Carnegie Airborne Observatory (CAO) light detection and ranging (lidar) system to create a comprehensive map of aboveground biomass (AGB) across the apparently mature ecosystem. Candidate explanatory variables were then developed using geology and parcel maps and a digital elevation model. Finally, multiple linear regression was used to develop a model to predict aboveground biomass from environmental data, and these results were compared to the lidar-based AGB estimates.

Results/Conclusions   To project AGB from the field data to the entire site, we tested many lidar-based metrics, but found that average height per plot provided the most successful metric across all vegetation types (observed v. predicted r2 = 0.71). We also found that though insolation and substrate were strong predictors of AGB, historic ownership was also important. Surprisingly, the final model only explained 44% of the variation in AGB. Analysis of the uncertainty map and historical photographs show that in many areas of poor fit, either small disturbances took place in the first half of the 20th century or major hydrological changes are currently underway. These results indicate that other processes, like unmapped land use history and community and population dynamics, are still important drivers in this ecosystem, despite its apparent steady state. Furthermore, these results suggest that in efforts to use “reference ecosystems” for conservation and restoration planning, care should be taken to consider how much of the observed heterogeneity is physically based versus how much is a consequence of temporal processes that may be very difficult to replicate.