why we transform our data
Flight data for Scirtothrips dorsalis around two hosts, Rosa 'Radrazz' and Conocarpus erectus for approximately one year - before and after transformation by the natural log. Kindly ignore the posted trendlines, as they are mostly irrelevant and unduly influenced by start and endpoints.
BEFORE:
There are obviously absolutely more thrips flying around the roses than around the buttonwood year-round, but a more interesting question to ask is whether the rate changes between those notable peaks and valleys are similar for the two hosts. This graph suggests some similarity between the two flight populations, but scale makes it difficult to eyeball an answer. I am further plagued by inconstant variance that increases as a function of the mean - not an atypical trend among natural biological populations. This linear transformation I will apply helps to normalize the data, and it will also remove some of the problems of scale:
AFTER:
While I still need to go through and confirm statistically that these weekly rate changes line up between the two hosts, it does appear a lot more likely that thrips are changing their weekly flight behavior around both hosts in a similar fashion throughout the year. Host is certainly relevant to the total number of thrips available for flight (data not presented here!), but other factors which are constant for both hosts might explain the relative weekly rate of thrips in flight about those hosts.
I obviously need to compare this data to environmental factors, and to the on-plant population densities that I have also recorded during the last year.
Is this progress? I don't really know. It is a hell of a lot of data to poke through.