1 Topic by dave kilbey

  • dave kilbey
  • Member
  • Offline
  • Registered: 14-08-2012
  • Posts: 29

Topic: Recording nothing at all within the sample > occurrence model

For a project we are doing (Solent Birds) we would like to be able to record when no species are seen at a particular location.  In principle this is fine but Indicia doesn't really like this type of setup.  We have a survey + species list so recording a taxon (or all taxa) with no abundance is doable but recording no occurrence at all seems more complicated, particularly if we want to see the null surveys appear in the verification report.  We do have a verification report that works with Samples only but it complicates the client's situation more and their old data conforms with the model of having an occurrence/taxon that is effectively "none".  So far better that we have a way to see that person x went to location y at a certain date and time and recorded "no species present" as a single entity rather than a list of 50 bird species appearing all set to zero

Some ideas:

If I were to add a taxon called "none" to our Solent Birds species list can I do the following in something like the "sample with occurrences" form:

a) Selectively prevent it from appearing in forms where I've used the [species] control?
b) Conversely, only show the option for "none" (filter everything else out) so that we can provide it as an option by again using the [species] control in a form?

There's the Species "menu" which looks like it might be able to do the job but I couldn't get anything like "taxon filter items" to work.  It would be nice to have the ability to filter out all taxa from the list except "id=64" (and vice versa)

Another option would seem to be to have another species list specifically for recording "none" within the same survey.  Is that possible?  In this case, we could have a separate sample with occurrences form for recording "none" and wouldn't then need to play around with filtering the results of the [species] control.  But can you/how would you identify which species list the control should utilise (and would this approach cause any knock-on effects - I suspect it would in terms of us requiring some mechanism to include the new "list" in other reports)?

As a further small issue, adding it to the existing list of 50 species would presumably mean that all of our report summaries etc would update to show 51 species which is incorrect.  I assume that a mechanism exists to filter out the "none/no species present" option from them too does it?

Thanks

Dave

2 Reply by Jim Bacon

  • Jim Bacon
  • Veteran
  • Offline
  • From: BRC
  • Registered: 03-11-2008
  • Posts: 980

Re: Recording nothing at all within the sample > occurrence model

Hi Dave,

If I think about the Indicia data model, the logical way, in my mind, to represent a record where no species are seen would be a sample with no occurrences. I think that is entirely fine from the warehouse point of view and the sample with occurrences form.

I'll just note that a record with no species is different from a record where you want to record absence of a species which would be a sample with an occurrence of that species with the zero_abundance field set to true. You would probably only want to use that in cases where a thorough survey of a location has yielded no evidence of a species being present. I hardly need to say that failure to observe something is not the same as observing absence.

Regarding verification of a record where no species are seen, just what are you verifying? Verification is all about confirming an occurrence was correctly identified. You could easily create a report that showed who went where and how many occurrences they recorded where that number could be zero.

Adding a taxon of "none" might be one way to signify that nothing has been seen but it would create an occurrence where, logically, there should not be one. You could write JavaScript to alter the forms when that taxon is used and you could re-write reports to filter out records with this taxon. I think this is a bad way to go though.

Instead, consider a sample attribute which is a boolean for "No species seen". You can use this to trigger JavaScript to modify forms or to select/exclude records from reports without creating spurious occurrences.

Jim Bacon.

3 Reply by charliebarnes

  • charliebarnes
  • Veteran
  • Offline
  • From: Greater Lincolnshire
  • Registered: 07-09-2008
  • Posts: 1,384

Re: Recording nothing at all within the sample > occurrence model

We do as Jim says for one of our surveys - a sample with no occurrences (and we record additional 'sample' data regardless). The key is ensuring these empty samples are shown where they need to be.

Charlie Barnes
Information Officer
Greater Lincolnshire Nature Partnership

charliebarnes's Website

4 Reply by dave kilbey

  • dave kilbey
  • Member
  • Offline
  • Registered: 14-08-2012
  • Posts: 29

Re: Recording nothing at all within the sample > occurrence model

Thanks Jim

You'll be glad to know that what you have said mirrors what John has discussed with us prior to my latest "bright ideas" for matching what we're attempting to do with what the client has done before in a previous incarnation of the project.

The idea behind verification as I see it would twofold.  Firstly, it would be a way for the client to see all the records as they come in one place/report and also to give them the ability to say "yes, this is a correct record and not spam" or to verify that it is in a sensible location (not from the middle of the Atlantic for example).

I'll reflect this back to the project and see what they think.  It's difficult finding a solution that meets requirements and fits with the Indicia data model.  We'll get there eventually!

Thanks again,

Dave

5 Reply by dave kilbey

  • dave kilbey
  • Member
  • Offline
  • Registered: 14-08-2012
  • Posts: 29

Re: Recording nothing at all within the sample > occurrence model

Hi Charlie

Thanks for that, good to know.  When you say "the key is to ensure the empty samples are shown where they need to be" do you mean in terms of them appearing within reports that within Indicia tend to naturally centre around showing occurrences?  Do you have the records appear in verification reports?  How do you get to see them as they come into your system?  Have you written a report/s to show them specifically?

Thanks

Dave

6 Reply by charliebarnes

  • charliebarnes
  • Veteran
  • Offline
  • From: Greater Lincolnshire
  • Registered: 07-09-2008
  • Posts: 1,384

Re: Recording nothing at all within the sample > occurrence model

Yes in Indicia reports - although we are a custom use case (John designed us a form), so empty samples already show up for person doing data entry "naturally" (it's a list of species down the left, weeks across the top - aka cross tab).

However, in terms of verification/data use, empty samples aren't currently analysed so I'm not much help there I'm afraid - although verifiers can see the data entry forms and therefore the empty samples.

Charlie Barnes
Information Officer
Greater Lincolnshire Nature Partnership

charliebarnes's Website

7 Reply by Jim Bacon

  • Jim Bacon
  • Veteran
  • Offline
  • From: BRC
  • Registered: 03-11-2008
  • Posts: 980

Re: Recording nothing at all within the sample > occurrence model

Hi Dave

Indicia has been developed for species recording so verification revolves around occurrences.
From the point of view of species recording, if you saw nothing, there is nothing to record.
If you record a sample with no occurrences, does it matter whether the location is accurate or not? What is it going to be used for?

Does this project record lots of interesting sample attributes which make the samples interesting even when there are no occurrences? Is it very important to measure participation? Does it matter if there are some erroneous samples present?

In effect, you are asking Indicia to do a different job from the one it is designed for if you want to verify samples.
If you are doing something that no one else has yet needed to do, it may be worth examining why to make sure it is necessary.

Of course there is always a solution. Indicia is extensible so you could write a module to add e.g a record_status field to the sample table.
You could re-write the verification form to show a tree of samples (sub-samples) and occurrences and allow you to verify at any level.
You could cascade rejection so a rejected sample would have all its occurrences rejected.
You could rewrite reports to consider sample.record_status when selecting records.

Is it worth all that work?

You could just delete a record that is spam. It's not exactly friendly but neither is spam.

Jim Bacon.

8 Reply by Jim Bacon

  • Jim Bacon
  • Veteran
  • Offline
  • From: BRC
  • Registered: 03-11-2008
  • Posts: 980

Re: Recording nothing at all within the sample > occurrence model

Hi again,

You could always add a custom sample attribute for record_status which is not shown to recorders and create a prebuilt_form for listing samples which has verify/reject buttons to set the sample attribute.
Your reports would have to take account of the attribute value.

Clicking reject could also set associated occurrences to rejected but clicking verify could maybe link to the verification page and list the occurrences of that sample.

Jim Bacon.