Hi. I am trying to help provide some rough advice for selecting amongst alternative sampling designs to measure bird density or abundance. The surveys are short-term in duration such as for impact assessments and habitat association studies. In this instance, I am looking at rough guidelines for how to select where to sample (survey design) and not how to sample (survey method). Attached is a rough first draft of a decision tree (see attachment) to select between various forms of study design ranging from conducting a census to spatially balanced sampling & stratified random designs. Do you have thoughts and suggestions? Is neglecting to include simple random sampling a fatal flaw for example? If so, where would you place it? Of course, the main advice will be to consult with a biostatistician, but hopefully this (with come accompanying text and references) can provide some rough guideance and be a starting point for that conversation. Literature, debate and suggestions welcome and appreciated.
The best way is to walk a transect regularly during different times of the year and calculate counts or densities. You don't have to do it randomly because you might be interested in a particular area. And what if your random design causes you to sample where there are naturally no birds? If there isn't a bird, you can't count it lol
@Andrew Paul McKenzie Pegman. Thanks for your response. I disagree though. Taking a randomized sample to generate estimates allows valid extrapolation to the full project area. That includes predicting where a species does NOT occur. That can be very important, for example in the context of surveys for a rare species in an area slated for potential development. Quantitatively showing where the species does not occur can provide the method of mitigating effects (I.e. place the development pr disturbance where the species does not occur and avoid where it does).
Steven L. Van Wilgenburg I agree with your last post. Have you tried searching the relevant literature?. If I were doing it I would do it as I advised in a recent post. Divide your total area up into strata and take a simple random sample from each stratum. My post here should help:
https://www.researchgate.net/post/What_should_I_do_to_determine_the_sample_size_per_hospital If you have further questions please ask. Best, David Booth
PS I would make the strata small even though that increases labor. then you would have less trouble with your results. I might even try for a census in each stratum to get the best possible results. remember a census is a sample with full information at the time you took it. BTW this link may be helpful: https://www.google.com/search?q=Sampling+wild+bird+populations&rlz=1C1CHBF_enUS874US874&oq=Sampling+wild+bird+populations&aqs=chrome..69i57.20095j0j1&sourceid=chrome&ie=UTF-8
- Badges
- Science method
- Similar topics
- Mathematical Sciences
- Statistics
- Applied Statistics
- Survey Design