I would like to introduce a proposed method of rapidly surveying vegetation using iNaturalist and citizen science. Here is a quick rundown of the method:
Firstly, a number of sites within a given vegetation type is selected randomly (~20). In each of these sites, the first ten plants observed are recorded and a picture is taken. This generates a list of dominant species in the area, as well as a proxy for abundance (the number of times a species pops up in the ‘ten’ species of each site list.) The pictures of each plant can be uploaded to iNaturalist, with a site tag and GPS location, where it can be identified.
The assumed advantages of this simple method is that it takes far less time than conventional, quadrat/cover abundance estimation sampling, it also can be done by ‘parabotanists’, and it can be conducted across vastly different vegetation types. This may only give the very basic elements of a community but may be appropriate for vegetation mapping.
I would like to know what the forum thinks of this, how feasible could it be to sample in this manner, how appropriate would it be to use iNaturalist for this, and what possible drawbacks might occur.
Also, If anybody knows of a similiar method that has been used elsewhere, please let me know as well.
Not a botanist, but the only major issue I see with this is in the selection of the first 10 plants. I would guess that surveyors are going to be likely to choose bigger/more boldly colored plants that stand out than others rather than drab ones. Otherwise, this type of protocol would likely do a reasonable job of characterizing the most common plants in an area I would think.
I don’t see any problem with uploading observations like this to iNat. The usual caveats about making sure the observers are making good quality observations with all appropriate info and pictures of key characters that allow for most things to be IDed would apply, but as long as the observers follow the rules of iNat, post away!
The issue you brought up was the first possible problem myself, and some other people thought of too. One thing I forgot to mention is that these sites will encompass as small as possible area, the sampler would preferably stay on the same spot while documenting these ten species. In many cases, this would force to the sampler to look beyond the first 4/5 charismatic species in the immediate area. I think an ideal cut-off distance for the sampler to move would be about 3m either way. In the few practice runs I have conducted I found myself needing to really look close and get on the ground to find the last three or four species.
Of course, the bias may still arise even in this case, so another option would be to tackle this directly and offer a sort of training session to bring this problem up to potential samplers?
The area you need to look at to get a representative sample of species depends on the vegetation type, and within that on the sizes and distributions of individual species. Typically, to describe vegetation an ecologist would first subdivide by canopy layers – very simply for example: trees, shrubs, forbs/grasses, mosses/lichens. In a forest or woodland, you might record all the trees in the entire stand, then shrubs in smaller plots, then forbs etc. in even smaller plots. In a meadow, recording ALL the species in quasi-randomly placed 1 m square quadrats would give you +/- unbiased species frequencies (you’d have to count individuals to get actual abundance). It all depends on your purposes - just walking around and noting the visually dominant species can give you enough for a general vegetation classification if you’re not worried about fine details. If you can describe in more detail what you’d like to measure in your study, we can probably suggest specific methods.
The issue I see with it is a different one. The first ten plants seen are likely to be the ones nearest the entrance to the site. So unless the “site” chosen is a set of coordinates, that could bias the sample. And now I see that @abdullateefismail has addressed that point.
Another issue I see is in forest: the first ten plants encountered might be understory herbs and shrubs, before the observer gets to the first tree.
The primary purpose of this method is to quickly ‘ground truth’ vegetation mapping. Here, the thinking is that an analysis of the ten species in each site of a given area will provide enough information to tell whether a particular location is within grassland or forest etc. It may also relay the vegetation type it occurs in (grassy fynbos vs montane grassland for example).
After doing some preliminary surveys, the problem with scale was really apparent. For communities composed of larger individuals, like forest, there will definitely be a need to increase the size of the ‘plot’. Thank you, I appreciate the feedback.