![[logo] US EPA](http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Define Data Handling & Analysis Protocols
- Classify coral reefs
- Develop testable hypotheses
- Define data collection protocols
- Biological sampling
- Screen attributes to define metrics
- Determine appropriate sampling effort
- Define data protocols
- Validate decision processes
- Define biocriteria
- Implement monitoring programs
- Diagnose causes of degradation
- Evaluate management effectiveness
- Communicate results
Test repeatability of the protocol.
Repeatability means that repeat applications of the protocol will tend to yield similar values. For coral reefs, repeat measurements of corals made at a reef station should yield similar values for coral metrics. If not, the protocol may need to be modified to collect data from a larger area, or additional training may be needed for dive crews to ensure similar methods, or replicate samples may be required to estimate reef condition based on average metric values.
Conduct a statistical power analysis of reef metrics.
Figure. Repeat coral sampling of the radial transect to test repeatability of the method as applied by USVI divers (foreground) and EPA divers (background).
Statistical power is defined as the probability of detecting a change when a change truly occurs. Power is a function of the variance, the more variable an indicator the more difficult it will be to reliably detect a change in that indicator's values.
If 10 reef stations are visited repeatedly over time (paired test) a change in reef surface area from 45 m2 to ~22 m2 or less would indicate a statistically significant decline. For a 2-sample test in which different stations are visited each year, a much larger change would have to be observed to indicate a significant change. For 20 stations, a decline
Figure. Power curves for coral metrics. Upper green line is the mean value observed for reef stations in the Florida Keys
% Living coral was a more sensitive indicator for detecting change than total reef surface area. For repeat visits to the same 10 reef stations, a decline of ~10% or more would represent a statistically significant change in reef condition.
Lower lines indicate the mean value that would represent a statistically significant decline if 10, 20, 30 or 50 reef stations were sampled. The solid line represents a paired test in which the same reef stations are sampled each year; the dashed line represents a 2-sample test in which different stations are sampled each year.
Define Quality Assurance and Quality Control procedures.
Data handling protocols are best defined before data are collected. Loss and corruption of data can easily occur when data are collected over many years, as is the intention for most long-term monitoring programs.
Original data should always be archived. New data should be saved along with data collected in previous years in the same files and in the same format. Documentation describing variable names, units, and collection methods should always be saved as well.