|
Activity Number:
|
671
- Environmental and Ecological Monitoring
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 3, 2017 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
|
Abstract #323521
|
View Presentation
|
|
Title:
|
Statistical Design and Analysis for Plant Cover Studies with Multiple Sources of Observation Errors
|
|
Author(s):
|
Wilson Wright* and Kathi Irvine and Jeffrey Warren and Jenny Barnett
|
|
Companies:
|
US Geological Survey and US Geological Survey and US Fish and Wildlife Service and US Fish and Wildlife Service
|
|
Keywords:
|
beta regression ;
imperfect detection ;
observer errors ;
plant cover
|
|
Abstract:
|
Field studies have documented potential observation errors in visually estimated plant cover include recorded measurements which differ from the true value (measurement error) and incorrectly recording a species is absent within a plot (detection error). Zero-augmented beta regression is the most suitable method for analyzing plant cover recorded as a proportion or percentage areal coverage of a plot but assumes no observation errors are made. Using simulations, we explore how inferences are impacted when data are analyzed with zero-augmented beta approaches that ignore observation errors. For comparison, we present a Bayesian hierarchical extension that explicitly models the observation process thereby accounting for both measurement and detection errors. We found the magnitude of bias using approaches that ignore observation errors was related to the degree of imperfect detection. Explicitly modeling the observation process within a hierarchical framework produced unbiased estimates and nominal coverage of model parameters. We also explore the effect of sample size and within season revisit design on the ability to detect a change in mean plant cover using our model.
|
Authors who are presenting talks have a * after their name.
