Optimizing foliar chemistry subsampling methods

Optimizing Foliar Chemistry Subsampling Methods

1.  Background/Question/Methods

Over the next 30 years the National Ecological Observatory Network (NEON) will monitor environmental and ecological change throughout North America. When fully operational, NEON will provide a suite of standardized data from several ecological areas including biomass and biogeochemistry, from 60+ sites across 20 eco-climatic domains. The breadth of this sampling regime, particularly for foliar chemistry, provides a comprehensive illustration of biomass and nutrient cycling within sites, but also poses budgetary and logistical challenges. The collection and laboratory analysis of foliar samples is time consuming, costly and in some cases extremely challenging, depending upon canopy height and density. Here we examine the adequacy of subsampling foliar chemistry from a large number of samples collected during prototype efforts.  The primary goal is to optimize both sample size and number, such that standardized uncertainty limits may be achieved with a minimum amount of sampling effort. In order to mitigate budgetary and logistical constraints, NEON will rely on several sampling techniques including airborne data collection, remote sensing data, and foliar chemistry sampling to capture a holistic picture of foliar biomass and biogeochemistry at each NEON site.

2.  Results/Conclusions

In 2014, NEON collected foliar samples from an oak and pine savannah at two sites in NEON’s Southeast Domain (D03) – Ordway Swisher Biological Station in Florida and Jones Ecological Research Center in Georgia. Using an extendable pole trimmer, leaf samples of dominant and co-dominant species were collected from trees located within 40x40m sampling plots.  Sub-samples were analyzed for leaf mass per area (LMA), stable isotopes of C and N, and percent weight of C and N.  This ground-based data allows for the monitoring of changes in canopy nutrients over time, and can be linked with airborne LiDAR and spectral measurements to map canopy N.  To estimate the mean sample size to within a standardized level of uncertainty, we ran regression analysis on the LMA values by species of subsampled individuals.  We also employed a Semivariogram approach to determine ideal sample size.  By utilizing this prototype data to run the above analysis, we are able to inform the optimal dataset size across sites to maximize returns within future financial and logistical constraints encountered by NEON.