I am trying to study selection of post-foraging perching tree by frugivorous birds.
I have collected perching frequency of bird used tree, and also studied of the number of available trees.
Does anybody have any good suggestions for evaluate tree selection?
To study tree selection with your data I will apply Chi square tests comparing the observed with expected (supponsing no preference for any sutdied tree) theoretical data. Details of these methods could be found in many book of general statistical methods.
Ning Li -
This is not my area, but I am curious about it. Are you saying that you have a frequency of sightings of given birds in trees in an area, and want to know why they were selected by the birds? If so, could you take a map of the area, divide it into grids, and mark the grids by frequency of use in all trees in each grid? The grids should have the same number of trees in each, and be of the same geographic area, so maybe that would not make sense. Do you have trees picked in a uniform distribution across the area so that this might make sense? Are trees of comparable size? Perhaps then the grid blocks could be shaded from white to black, with increasingly darker shades of gray in-between, and if geographic features, like rivers, lakes, etc., are shown, this might tell you something?
Or are you looking at individual trees and might want to rank them by frequency?
I'm just not clear on your goal, though it may be obvious to people in your discipline.
Just curious.
Cheers - Jim
PS -
Perhaps you might find something of use among the papers of the following two soil scientists I noticed on ResearchGate, if my guess is correct about the kind of thing you are doing, but it is just a guess:
https://www.researchgate.net/profile/Jaap_De_Gruijter2
https://www.researchgate.net/profile/Dick_Brus
I refer to papers they did in spatial analysis/geospatial analysis.
I've done some selectivity studies (habitat use and diet). My first suggestion is to avoid simply dividing use by availability as this ratio is unbound and biased to small sample sizes and either extreme.i use Jacobs index (based on Jurgen Jacobs Oikos paper from the 1970s. This and Strauss 1974 highlight why their methods are better than straight ratios. If you have data on lots of individual birds, then you can calculate a Jacobs index value for each habitat used by each individual and then test these against a mean of 0 using t-tests. If not, the chi squared method (as described byJuan) or compositional analysis can be used.
Cheers
Matt
Well, if you use hypothesis tests, you should be careful not to misinterpret the p-value. They require a type II error analysis or similar sensitivity analysis for decision making. (Effect size is important to be practical, and standard errors are fairly interpretable, but I don't know what applies to your application.)
https://www.researchgate.net/publication/262971440_Practical_Interpretation_of_Hypothesis_Tests_-_letter_to_the_editor_-_TAS
I'm not familiar with your application, but if you use hypothesis tests, be careful you are not misinterpreting them.
Article Practical Interpretation of Hypothesis Tests - letter to the...
Perching trees, by definition, are species that are not used for foraging. Perhaps just report which species are used, and their frequency, because this is interesting in itself. I've reported some data on this in my latest paper, 'Size-based fruit selection by a keystone avian frugivore', available soon on this site and in the New Zealand Journal of Botany.
Andrew :-)
- Badges
- Science topic
- Similar topics
- Biological Science
- Zoology
- Ornithology