COS 113-4 - Optimizing monitoring strategies in the semi-arid shrublands of southern California: A variance decomposition approach

Thursday, August 6, 2009: 2:30 PM
Grand Pavillion V, Hyatt
Spring L. Strahm, Biology, San Diego State University, San Diego, CA and Douglas H. Deutschman, Biology Department, San Diego State University, San Diego, CA
Background/Question/Methods

 Monitoring to detect ecological change is an important component of many conservation programs and in southern California’s multiple habitat conservation areas monitoring is required.  Monitoring a large network of land is scientifically and logistically challenging as well as costly. The objective of this project is to evaluate the cost and accuracy of different sampling designs and field protocols for monitoring coastal sage scrub (CSS) and chaparral vegetation communities by evaluating sources of variability. We consider three major sources of variation: temporal (interannual), spatial and methodological. Spatial variation includes three nested, progressively finer levels.  Methodological variation includes two levels: protocol and team. Several suites of response variables were analyzed including species richness, cover of major functional groups (e.g. native shrubs, non-native forbs), and several example species from each functional group.  In 2007 and 2008 we sampled CSS and chaparral in San Diego, Orange and Riverside counties contained inside conservation lands.  Plots were sampled using three different protocols (visual estimation (2007 only), point-intercept transects, and 1m2 quadrat estimation), and were sampled by multiple teams.  Cost was recorded in terms of time.  Response variables were analyzed using an additive GLM, and the change in variance explained by the addition of each new variable was recorded. 

Results/Conclusions

 Semi-arid shrublands in southern California are highly spatial, with different species and functional groups displaying different degrees of patchiness across multiple scales.  As a result allocating a significant amount of effort to spatial coverage is appropriate for most response variables.  Some species and groups are also dramatically influenced by annual factors such as rainfall, and require annual monitoring, where as others can be sampled less frequently.  Team to team variability can be minimized with appropriate training and experience.  Transects provide the most accurate and precise estimates of cover for individual species and functional groups. Quadrats provide more information on richness and presence of uncommon or small species, but systematically underestimate cover.   Visual cover estimates do not add any additional information when the other two techniques are applied.  Our data demonstrate that response variables vary across natural spatial gradients and temporal variability, and that different field protocols capture different aspects of the ecosystem. The best monitoring approach must be determined based on the objective(s) and response variable(s) of interest for each individual project.  Variance components analysis can provide important insight on how to allocate effort across different sources of variability once the scope of a project is understood.

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