PS 38-95 - Long-term trends in plantation forestry: A remote sensing approach

Wednesday, August 9, 2017
Exhibit Hall, Oregon Convention Center
Devin E. McMahon, Earth System Science, Stanford University, Stanford, CA and Robert B. Jackson, Stanford Woods Institute for the Environment, Stanford University, Stanford, CA

Plantation forestry, in which trees are grown as a crop, may reduce global deforestation if plantations can continually produce high wood yields on a limited land footprint. However, exceptionally rapid growth and efficient nutrient mobilization by plantation trees, combined with soil compaction and removal of nutrient-rich plant tissues in harvest, can alter soil physical and chemical properties and limit future growth of plantations or subsequent native vegetation. To investigate the long-term sustainability of plantation forestry, we are studying the highly productive commercial eucalyptus plantations of southeastern Brazil.

This work uses multi-decadal time series of vegetation indices to track aboveground productivity over multiple harvests in thousands of individual plantation units. For each plantation, we segment a time series of seasonal maximum NDVI values extracted from 32 years of Landsat imagery, in order to identify the planting and harvest dates of each rotation (harvest cycle). We then use best subsets regression methods to identify environmental variables, such as annual precipitation, that may drive different trends in vegetation index values over successive rotations. Available field data show that our method is able to identify the starts of harvest cycles with approximately 85% accuracy, and we are assembling a biomass dataset to identify the vegetation index-based metrics most strongly correlated with plantations’ aboveground productivity over time and space.


Preliminary results from a subset of the study area show substantial increases in vegetation index values between the first and second rotations since 1984, consistent with management improvements, followed by a plateau in rotation-average VI values, with values peaking around the fourth of four to five rotations. Our results do not strongly support practitioners’ anecdotal reports of widespread productivity declines with successive rotations. However, we do observe decreasing rotation lengths over time, suggesting further acceleration of nutrient export. Soil type and elevation, rather than precipitation received over the course of the rotation, are most strongly correlated with different plantation productivity trends across the study area, supporting the hypothesis that soil processes control plantation sustainability. Ongoing work will incorporate additional plantations across a broader precipitation gradient, and we will collect soil nutrient data from across the study region to further test this hypothesis.