![[logo] US EPA](http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Trend Assessment
Description: Trend may refer to the changes observed at a single site, or to changes through time for a group of sites within a region. If trend sites are selected randomly, the changes observed at those sites can be generalized to the larger region. If sites are selected using any other method, the results can only be used to describe those particular sites.
Simple example: The riparian habitat surrounding an urban stream is being restored and you want to test whether the invertebrates indicate an upward trend in stream condition. You select a site and sample the invertebrates from that location every year. You calculate a multimetric index and test whether the values goes up over time. To do this, you would regress index values against year.
MAIA example: The original MAIA project did not encompass a long enough time period to test for trends. However, repeat visits to a subset of sites allowed us to calculate the amount of change through time that could be detected for different monitoring designs. The statistical model used was regression of index values against year.
Figure 1
Figure 1: The percentage change per year that would represent a statistically significant change in multimetric index values decreases as the number of years of sampling increases. A smaller percentage change indicates a more precise index and greater statistical power to detect change. Thus, the invertebrate index was the least precise although all three indexes were very similar. Shown are results for the same 40 sites visited each year.
How the method works: The most efficient way to detect a trend through time is to visit the same set of sites repeatedly. Revisiting the same sites eliminates the variability associated with different site locations because every site is compared to itself through time. Rather than sample every year, a more efficient approach divides the trend sites into groups, or panels, and might visit each panel of sites every five years in rotation. If the sampling sites are randomly selected during the first year, any trends observed represent changes in the region.
Figure 2
Figure 2: Example of a rotating panel design with 10 sampling sites located in each of 5 management units (e.g., watersheds) of a region (e.g., a county or state). The same sites in each panel are sampled every 5 years.