Abundance can be expressed as total, mean or as a proportion or percentage. For correlating with habitat variables in the absence of temporal data, which do you think is most applicable ecologically and also for testing trapping methods.
It is possible, you can try using richness and abundance as response variables in a GLM with the correct error distribution, it could be Poisson or Negative binomial. The problem is the quality of your environmental data, but you can do it, for example, to see if abundance is explained by temperature.
Hi Todd, you can use whatever measure of relative abundance you decide best represents the data you have. It is more a matter of consistency than choice of metric. It is also important to ensure that your results do not overstate the abundance of certain species over others for the area/habitats you are surveying.
The example I often used is for our bird surveys. We know we have species that are highly territorial and relatively cryptic, and so these are generally under surveyed. There are species where individuals will tend to follow the observer around and it is easy to inflate their abundance through "double-counting" (on more than one instance). Transient species present a problem in that they are obviously utilising the habitat but move through on an irregular basis, so what would be a realist abundance for them?
To resolve that issue we will utilise the max number observed on any one day for that species, e.g.;
Day 1 2 3 4 5 6 Relative Abundance
Sp1 (transient) 0 20 4 0 38 0 38
Sp2 (resident) 1 2 2 2 1 1 2
Sp3 (cryptic) 0 1 0 0 0 1 1
etc
It gets a little easier for mammals as individuals can be marked and so you can get a better estimate of the numbers. Reptiles can be marked as well but previous studies in the habitats we work would suggest a tally for each species captured/recorded during the survey is likely to be a slight underestimate.
Once you resolve the issue of abundance measure, the analysis for environmental variables is relatively simple (as Pedro suggested).
Your qualifier that you do not have temporal data can be an issue if that relates to comparing environmental variables and survey observations from different periods - seasonal variation may give you misleading outcomes from the analysis.
"Relative abundance" is ambiguous in this question. It is often employed for an index (see Johnson, D.H. 2008. In defense of indices: the case of bird surveys. Journal of Wildlife Management 72(4):857–868), in which case the results cannot be compared across species. The key point with an index measure is that you assume that the method used is consistent enough that the index approximates a multiple of the true (but unknown) actual number or density. Factors that may not be controlled and that can affect the results can lead to detectability being confounded with number. For example, if you are interested in the difference between two "habitats" (biotopes) where detectability differs by a substantial amount, you are in trouble. This is also the reason why such methods usually cannot compare species, as each species has its own individual detectability and why it would be dangerous to ignore seasonal differences in detectability. I agree with most of what Eddy says, but not with his advocacy of using maxima, because such measures discard most of the information that you have obtained and so deliver less precise measures. It's best to use some measure of central tendency, for example arithmetric or geometric mean.
Hi David. The methods we employ is supported by 35 years of monitoring data (over 25,000 data points). It was and is never intended to provide a measure of population sizes. It is simply a relative measure describing biotic assemblages within habitats. It is not meant to describe individual species to any great depth because the data, in that format, is not designed to do so. If we wish to investigate species or groups of species (guilds) then we can drill down and determine the appropriate measure of abundance in those situations.
As indicated we have used the system to explore and assess rehabilitation over a few decades and that has facilitated and directed changes in processes to improve and enhance rehabilitation of habitats.
When investigating particular species groups, we take a different approach to the data and how we use it. More importantly, we assess the data and ensure that we do not apply analytical processes that are more fad than reason. Our guiding focus is that the data dictates your analysis, not the analysis dictating your data. But as always, unless you have an understanding of the species and systems you are studying the interpretation of the data can be misplaced.
To be honest, my own experience is there is a tendency for researchers, especially those who lack the on-ground observational experience, to over-analyse data and come out with very curious results that are not backed up by observation.
Eddy, I have more years of experience and number of data points than you, but this proves nothing. I don't know what you mean by "biotic assemblages", but any measures you may use will be affected by detectability and may therefore reflect details of methodology and detectability as much as actuality. I suggest that you read Johnson, including other papers. He has many years of sound practical experience in advising on study design and has an excellent approach to the strengths and weaknesses of index methods. Beware measures of "abundance" (absolute numbers or density), most have strong asssumptions that are not met in practice. Indeed most are little more than elaborate, inefficient index methods. But we may be off the topic, which seems to involve trapping and for which the subject question and species were not specified.
David Dawson, I'm not quite sure what you are trying to prove. The statement that our methods are backed up by the data is unequivocal. We have studied an area for 35 years and collected over 25,000 data points and the analysis illustrates that our use of maxima for the bird survey data represents the best reflection of the biotic assemblages (i.e., the species list) for those habitats. It is a data driven outcome. It may not fit with your views but that is a moot point as we are trying to assist Mr Johnson. You are welcome to provide yours as you have. All I did was provide you with a basis for doing what we do in the subsequent response.
I stand by my original statement of methods.
PS this is not the forum to vent your issues. Please message me directly if you wish to continue.
In fact i compared the habitat variables with the population abundance/ density of some of the perching birds. I collected data on perch types, perch ht, perching ht, foraging ht, foraging distance from water bodies, foraging method etc, It is possible to get better results and interpretation by applying discriminant function analysis/ Principal component analysis rather then by simple correlation. .
I am grateful for your suggestions and comments. I now have an idea of what kind of analyses to use for the different trapping methods. Thank you once again to those that responded to my question.
- Badges
- Science topic