OPS 1-6 - A scalable strategy for plant biodiversity sampling

Monday, August 8, 2011
Paul Duffy1, David T. Barnett2, Elena Azuaje2, Rebecca Hufft Kao3 and Joel McCorkel4, (1)Neptune and Company, Inc., Bellvue, CO, (2)National Ecological Observatory Network (NEON), Boulder, CO, (3)National Ecological Observatory Network (NEON, Inc.), Boulder, CO, (4)NEON Inc., Boulder, CO
Background/Question/Methods

The National Ecological Observatory Network (NEON) will provide a physical and information infrastructure to enable understanding and forecasting of the impacts of climate change, land use change, and invasive species on continental-scale ecology over decades.  Understanding effects as described by terrestrial trends requires rigorous quantification at local sites.  NEON will design a system for sampling the diversity and abundance of biological organisms that is consistent with the goals of the continental monitoring network. 

The local study design for terrestrial sampling will mirror the NEON approach for objectively representing ecological variability at the national scale.  The United States was stratified into twenty relatively homogenous domains to increase sampilng efficiency and decrease sampling error.  Study sites were selected to be representative of dominant domain vegetation, soil, climate, and ecosystem performance.  At a minimum, a site-specific study design was created to quantify patterns in vegetation type. 

Results/Conclusions

The sites were selected to represent the following characterisitcs:

-  The sampling footprint was stratified by vegetation, and sample sites within each strata were distributed according to a spatially balanced system that was implemented in a geographic information system.  This will allow inference to entire populations and extrapolation to unsampled space.  

-  Measurements of selected taxanomic groups will be coordinated and co-located to describe co-variation across space and through time. 

-  The power analysis used to inform the computation of a minimum sample size was based on a repeated measure design within a linear mixed model.  The comparison of rates of change in an ecological response through time and between sites was considered as an initial case.  In the repeated measures framework, sample size is dependent on temporal variation at each site. Peak greenness was used as a proxy for correlation structure through time, and to determine that a compound symmetric correlation structure could be applied.  Further adjustments to stabilize the magnitude and variation of a difference in slopes between sites within a fixed period of time resulted in a minimum sample size that will be associated with estimates of population variance that vary according to each target taxon.

-  To determine sample allocation among strata, area within a stratum was used as a proxy for within stratum population variance. Under this assumption, sample allocation according to the square root of the area associated within each stratum should result in sample standard deviations that will be roughly equal among the vegetation strata within and across sites.

Copyright © . All rights reserved.
Banner photo by Flickr user greg westfall.