how to use isAbsenceReported and absentTaxa

[ymgan]

Is absentTaxa intended to be used for target taxa only? Or is it intended for non-target taxa?

Example 1

Is it correct to list target taxa that were not detected in absentTaxa?
Is it correct to say isAbsenceReported is true when target taxa not detected in the event is listed in absentTaxa?

Humboldt table

eventID	targetTaxonomicScope	absentTaxa	isAbsenceReported	isTaxonomicScopeFullyReported
EVENT_001	Nototheniidae | Myctophidae | Paralepididae	Myctophidae | Paralepididae	true	true

Occurrence table

eventID	family	occurrenceStatus
EVENT_001	Nototheniidae	present

Example 2

It is supposed to mean the same thing as Example 1, except the non-detection of target taxa is listed as per record in Occurrence table.

Humboldt table

eventID	targetTaxonomicScope	isAbsenceReported	isTaxonomicScopeFullyReported
EVENT_001	Nototheniidae | Myctophidae | Paralepididae	true	true

Occurrence table

eventID	family	occurrenceStatus
EVENT_001	Nototheniidae	present
EVENT_001	Myctophidae	absent
EVENT_001	Paralepididae	absent

Example 3

With this example, I believe I can infer that Myctophidae and Paralepididae were not detected in EVENT_001 (i.e. Example 1 and 2) because isTaxonomicScopeFullyReported = true:

Humboldt table

eventID	targetTaxonomicScope	isAbsenceReported	isTaxonomicScopeFullyReported
EVENT_001	Nototheniidae | Myctophidae | Paralepididae	false	true

Occurrence table

eventID	family	occurrenceStatus
EVENT_001	Nototheniidae	present

Do I understand correctly that the 3 examples above meant the same thing? Is there a preference of which one to use? I have this question when I am re-mapping our dataset ...

[kingenloff]

I think eco:absentTaxa could only really be used in the context of targeted taxonomic scope. We can only assert that an organism is absent/not detected if we were purposefully looking for it to begin with. Absence cannot/should not be asserted for bycatch.

Perhaps we should add this clarification as a comment for the term.

[rdstevenson10]

Good question, ymgan. The three options you outline seem logically identical. I am not sure which one is preferred for easy of use or accuracy. Question does the present list of terms have a boolean to say if the taxonomic scope is define?

Regarding kingenhoff's comment about bycatch or incidental catch. I generally agreed but here is a counter example.

Bycatch is common with some fisheries methods and can have important economic and legal consequences for commercial and recreational fishing. If one has the method and effort data for the bycatch and subsequently defined the detection with the method, one could infer absence later one. For instance in the case of catching marine turtles with fishing nets it could be an important conservation analysis. I see this as a plus.

One might then need an indicator for the cccurrence table saying there are observations included outside of the defined taxonomic scope.

Another approach is to do what I think eBird does. They have the TaxonomicScore defined as Birds implicitly. Then they have an occurrence table and finally they ask if the occurrence table is complete. This suggestions a new term isOccurrenceTableComplete. The eBIrd approach avoids the idea of absence for collection purposes but it become critical in analyses. "Absence" is a difficult concept for several reasons. One option would be to replace "absence" in the Humboldt terms with "non-detection" when reporting survey data.

I do not see that present eBird approach as being able to deal with "bycatch" except by putting an observation in the comments. The current ebird model makes it hard to report squirrels at a feeder or snake eating bird eggs.

[yanisica]

While discussing this issue with @ymgan we realized that it is not clear if the eco:absentTaxa refers to 'non-detections' or 'real absences'. Possible interpretations:

• eco:absentTaxa refers to 'real absences' (probably as the result of some analysis the data provider did) because 'non-detections' are captured in dwc:occurrenceStatus as 'absent'.
• eco:absentTaxa refers to 'non-detections' because it is the outcome of a survey (not the result of an analysis). If this interpretation is correct then eco:absentTaxa is duplicated with dwc:occurrenceStatus = 'absent'.

@tucotuco @kingenloff @baskaufs what do you think?

Similarly, is eco:isAbsenceReported only linked to eco:absentTaxa? or is it also linked to dwc:occurrenceStatus? (e.g., if a taxon has dwc:occurrenceStatus = 'absent' then eco:isAbsenceReported = true?)

[tucotuco]

My understanding is that real absences can not be treated in any way without analysis, and that the target of published data sets are primary data. If so, eco:absentTaxa can not refer to 'real' or 'probable absences' and must refer to 'non-detections'.

I do not think that eco:absentTaxa is a duplicate of dwc:occurrenceStatus. Why? One can publish inventory data sets without Occurrences and dwc:occurrenceStatus is an attribute of Occurrence, not of Event. That is, you don't have dwc:occurrenceStatus available in the Event Core and without eco:absentTaxa you would be required to have an Occurrence Extension to share 'non-detections'.

In the scenario where there is an Event Core, a Humboldt Extension, and an Occurrence extension, the non-detections can be more explicitly expressed than with an Event Core/Humboldt Extension combo alone. In the first scenario it can indeed seem that eco:absentTaxa and dwc:occurrenceStatus are redundant, but in fact they act on different levels Inventories versus their Occurrences.

Comments welcome.

[rdstevenson10]

Dear All I am trying to follow what you wrote, John. As I understand it, published data are primary data sets. In most of the literature I have read, “non-detected” would be a more precise term than “absence” for primary data, if the data are based on a survey or an inventory. Thus, I think I agree with what John wrote. I am not sure I follow the rest of your comment about events, occurrences, and the Humboldt extension (inventories) . Are you allowing for the fact that an event is hierarchical and could have both inventory data and occurrence data? Thanks, Rob From: John Wieczorek ***@***.***> Date: Friday, July 12, 2024 at 12:39 PM To: tdwg/hc ***@***.***> Cc: Robert Stevenson ***@***.***>, Comment ***@***.***> Subject: Re: [tdwg/hc] how to use isAbsenceReported and absentTaxa (Issue #103) CAUTION: EXTERNAL SENDER My understanding is that real absences can not be treated in any way without analysis, and that the target of published data sets are primary data. If so, eco:absentTaxa can not refer to 'real' or 'probable absences' and must refer to 'non-detections'. I do not think that eco:absentTaxa is a duplicate of dwc:occurrenceStatus. Why? One can publish inventory data sets without Occurrences and dwc:occurrenceStatus is an attribute of Occurrence, not of Event. That is, you don't have dwc:occurrenceStatus available in the Event Core and without eco:absentTaxa you would be required to have an Occurrence Extension to share 'non-detections'. In the scenario where there is an Event Core, a Humboldt Extension, and an Occurrence extension, the non-detections can be more explicitly expressed than with an Event Core/Humboldt Extension combo alone. In the first scenario it can indeed seem that eco:absentTaxa and dwc:occurrenceStatus are redundant, but in fact they act on different levels Inventories versus their Occurrences. Comments welcome. — Reply to this email directly, view it on GitHub<#103 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AGCKFMLVV7HWTNN27P3Q6YDZMABFLAVCNFSM6AAAAABIDRYGYGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMRVHE2DEMRXGE>. You are receiving this because you commented.Message ID: ***@***.***>