Researchers frequently compare rates of change based on counts (e.g. mortality, germination, infection, recruitment). The estimation of mortality and recruitment rates from tree census data have been discussed elsewhere. However, rather than estimating absolute rates, researchers may often need to determine whether two rates are significantly different. Most biometric texts and computer packages are not very helpful on this question, and guidance is not readily sifted out of the more technical literature. This note is intended to address three related issues: 1) to inform researchers who are not familiar with the appropriate tests, 2) to present an approach that avoids analytical complexity, and 3) to provide some feel for the power of the tests. This third point should assist in experimental design by providing guidance on the sample-sizes and differences required for significance to be detected. The simple and accurate graphical models provided will be of use to biologists who work with count-based changes. The number of observations required to detect significant rate differences may surprise some researchers.
Topic: change,forest trees,mortality,germination,diseases,recruitment
Publication Year: 1999
Source: Journal of Tropical Forest Science 11(1): 323-325