This was originally going to be just a descriptive table explaining how many species we sampled at each sample site relative to the total species pool sampled. But when I converted it to percent total species pool by site – there is an obvious visual trend. As you move from native prairie – through high-diversity restoration, low diversity plantings, and then ag field, the percent of the species pool declines. The raw species counts don’t lend themselves to analysis – but the percentages seem interesting.
This will be part of a paper that looks at insect community response to habitat restoration and this is not a critical analysis – but it’s interesting insomuch as it supports the general thesis of the paper – that you can create usable habitat for regionally imperiled insect communities via restoration.
I've attached both raw and % data
Comparing percentage across environments is the same as comparing raw data because you're just multiplying the raw numbers by the exact same factor. But it is useful to compare % WITHIN environments because it shows the proportion of each genera. But it is also useful to perform anova on the raw data with post-hoc tests to determine actual significant differences
I think you should do the analysis of the percentage to show a general trend. BUT don't forget: don't do the ANOVA directly on the percentages! you have to do 2 transformations of the percentage before: first calcul the square root of the percentage, and after the ARCSIN transformation of the square root. THen come the ANOVA analysis [see explainations in Zar 2009]. I would process a nested anova: highest level - types of pairies. Second level: the sites. Lowest level: the sampling points in each site. After you can calculate the percentage of impact / value of each 3 levels on the total variance (simple calcul well explained in Sokal and Rohlf 2012 ): so you will be informed the the type of prairie influence the variance of your data as XX%, the sites as YY% and the sampling points in each site RR% and discuss this information.
Interesting dataset regardless! I'm not sure that the % data help or hurt your case. If your data are normally distributed within each category, then stats on the percentages (after arcsine transformation) are hard to justify.
That should not stop you presenting the % data graphically if that makes the story visually more convenient. If you graph the raw data you run into issues with the less diverse taxa being flattened into the horizontal axis when graphed on the same scale as moths. Of course you could break the vertical axis and still make a useful graph with the raw data.
Fun dataset and a good problem to have!
- Badges
- Science topic
- Similar topics
- Biological Science
- Zoology
- Invertebrate Zoology
- Arthropods
- Insects
- Insect