How many pollen grains should we count? A basic statistical view

In pollen analysis, one question is often being asked: how many pollen grains should be counted for one sample? The proper count number should guarantee the reliability of the abundance (usually expressed as percentage) of the pollen taxa concerned in the sample. Practically, this number is usually more than 100 grains, but seldom over 1000. Most people tend to count at least 300 pollen grains (often of terrestrial plants)for each sample. However, this number is more likely based on the experience of earlier palynologists rather than on theoretical consideration. In this paper, we apply a simple statistical theory to evaluate the reliability of the observed pollen abundance with different total count number. The reliability is evaluated based on the error range. The results show that the reliability depends on both the true pollen abundance of the taxon concerned and the total count number: a low abundant pollen taxon needs a high count number to obtain a reliable data set. The conclusion is that there is no universal “proper” total count number for each sample. The total count number should be determined by both the abundance of the interested taxa and the research objective. The commonly used 300 grains may be good enough for most general vegetation reconstructions and for the accurate survey of pollen taxa with more than 10% abundance. For less abundant taxa, higher count numbers are suggested.