Visitation frequencies are derived entities and can be based on one or many actual flower visits (pollination events). For example observing 10 pollinator visits in 20 flowers over a 15 minutes period generate the same visitation frequency as 2 pollinator visits in 2 flowers over a 30 minutes period. In this example both observations generate the frequency of 2 visits per flower per hour. However, the first observation is based on the observation of 10 flower visits, while the second on only two. These realizations led us to investigate whether it would be possible to analyse all observed pollinators as separate events (counts) and using a Poisson or negative binomial error distribution, the distribution best fitting the type of count data flower visitation data represent. Trond Reitan has generated and analysed hundreds of thousands of datasets and we are happy to see that the paper is now in review in PlossOne.
Rethinking analyses of flower visitation data
During the analyses of a data set on flower visitations Dr. Nielsen and Trond Reitan started discussing the weaknesses associated with using visitation frequencies (i.e. number of visits per flower per time) as response variable in pollination studies. We have conducted a simulation study to prove our point.
Published Sep. 8, 2014 11:50 AM
- Last modified June 18, 2018 11:18 AM