Phylogenetic Justification text block not visible in calibration viewing page.

[Ksepka]

Can we have this block of text appear in results, just like the minimum age block of text does? It should appear above the tree graphic.

[jimallman]

OK, we're now showing the phylo-justification text for each fossil alongside the rest of its information.

But I gather you're asking for a proper heading like "node minimum age", then the justification text, but I don't think this fits the current data model. Each fossil has its own phylogenetic justification, and there can be more than 1 fossil per calibration.

EDIT: Just to be clear, each fossil has a phylogenetic justification that's unique to this calibration.

I've outlined the information for a single fossil on the calibration page (see the red dotted line). If we have a calibration with > 1 supporting fossil, the issue becomes more clear, since there will be a stack of these red boxes, one per fossil.

Suggestions are welcome.

[Ksepka]

There will not be more than one fossil per calibration in any cases. We ask authors to specify a single fossil if they try to list multiple, in keeping with best practices.

In fact, the reverse does occur - there are some fossils than are used in multiple calibrations. For example, Kimberella appears in five different calibrations. To date, the phylogenetic justification is always the same but in theory two authors could cite different justifications. This has not arisen yet and may never actually happen.

Given this relationship and the importance of Phylogenetic Justification, I think it should be given the same heading as Node Minimum Age.

[jimallman]

Can we confirm this with @pdpolly? The UI explicitly allows for multiple fossils for a calibration:

I'm pretty certain this was the result of extensive discussion, since it was not easy to make this a one-to-many relationship. That said, I don't see any multi-fossil calibrations in the current database.

Whether it's one or n fossils per calibration, I'd like to make the calibration editor and display page consistent about this.

[jimallman]

I found some notes and correspondence on this question. From an email from @pdpolly, Feb 1 2013:

[T]here could in principle be a many:many relationship between calibrations and fossils. An author might triangulate the age of a calibration using fossils on stem and daughter branches (many:1) or using several fossils from the same taxon that are from the same rock unit (many:1). Only one of these will be the oldest, of course, but the database would be richer if we record the entire suite of logic on which the calibrations are based. Conversely, two authors might use the same fossil in different ways, either because they have competing calibrations for the same node, or because the same fossil calibrates different nodes (e.g, an ancient ape fossil might be the oldest in Hominidae, but it might also be the oldest in the larger clade Anthropoidea).

I also have a note from the FC Working Group meeting at NESCent (May 5 2014), regarding the "total evidence" approach and occurence-based dating:

    > include all available occurence data
    > so why do we need min/max dates?
    ? Can FCD incorporate this (occurence data) in some way? [Rachel]
        ! yes, we already allow n fossils (occurences) [David]

[pdpolly]

The intention is to allow multiple fossils to be attached to a calibration. The many to one relationship allows for several possibilities: (1) cases where the author discusses the merits of more than one fossil, chooses one, but is uncertain enough to want to report all of them with the calibration; (2) cases where the author wants to list fossils associated with more than one daughter clade and/or nearest stem clade; and (3) cases where the author wants to provide the raw data that would be used to estimate the distribution of possible ages of the node using the density of fossils on branches above and below the node. In all cases the author still must indicate a preferred minimum age based on the criteria outlined in Parham principles.

[Ksepka]

My opinion after reading through this is that the Phylogenetic Justification actually fits best as a text field that is entered in association with the Calibration, rather than the Fossil Specimen. Is this possible, and if so do people agree with that idea.

My reasoning: as David notes, two authors may use the same fossil for different calibrations OR the same calibrated node but with a different Phylogenetic Justification text blurb (that actually has already happened - both Benton and I calibrate a node with Vegavis but are text is different since we wrote it independently. If the Phylogenetic Justification is attached to the fossil, then both Calibrations would need to use the same text.

A side note: I understand David's point about multiple fossils potentially being associate with a node. That said, in the ~100 calibrations entered so far, all of them use a single fossil in the database. E.G., I depict dozens of fossil occurrences in figures in my own FCS paper - such as the 20 or so Paleogene mousebird fossils that occur after the oldest record. This would allow interested readers to grab those data and use them to determine confidence intervals, etc. However, I only include the calibrating specimen in the FCD, which follows Best Practices. I would not expect the others to all to be entered into the FCD, which would take a mammoth amount of work (entering the fossil portion is the most time consuming of all steps).

[pdpolly]

Good point about the association of the phylogenetic justification. In data terms, the phylogenetic justification is a unique association between a particular node and a particular fossil. The node could have more than one fossil, and a phylogenetic justification would be needed for each, but a single fossil specimen could be associated with more than one node (or alternative calibrations for the same node). I can't remember how Jim decided to store the fossil specimen information relative to its association with node.

If I remember the discussions we had about possibility of multiple fossils, most people figured that each calibration would have only one. I think it was only Rachel and I who imagined a time in the near future when people might want to include more than one. It doesn't surprise me that all the members of the working group are using only one fossil, and that pattern may continue into the future.

On 9 Oct 2014, at 9:54 AM, Ksepka notifications@github.com wrote:

My opinion after reading through this is that the Phylogenetic Justification actually fits best as a text field that is entered in association with the Calibration, rather than the Fossil Specimen. Is this possible, and if so do people agree with that idea.

My reasoning: as David notes, two authors may use the same fossil for different calibrations OR the same calibrated node but with a different Phylogenetic Justification text blurb (that actually has already happened - both Benton and I calibrate a node with Vegavis but are text is different since we wrote it independently. If the Phylogenetic Justification is attached to the fossil, then both Calibrations would need to use the same text.

A side note: I understand David's point about multiple fossils potentially being associate with a node. That said, in the ~100 calibrations entered so far, all of them use a single fossil in the database. E.G., I depict dozens of fossil occurrences in figures in my own FCS paper - such as the 20 or so Paleogene mousebird fossils that occur after the oldest record. This would allow interested readers to grab those data and use them to determine confidence intervals, etc. However, I only include the calibrating specimen in the FCD, which follows Best Practices. I would not expect the others to all to be entered into the FCD, which would take a mammoth amount of work (entering the fossil portion is the most time consuming of all steps).

— Reply to this email directly or view it on GitHub.

[jimallman]

In data terms, the phylogenetic justification is a unique association between a particular node and a particular fossil.

This is how it is stored, in a table that links fossil A to calibrated node B. This allows multiple independent assessments of the same fossil. Fields in this table include:

• species
• fossil location relative to node
• min/max age and type
• whether dates are tied to a geo-time scale boundary
• phylogenetic justification and type- phylo publication

[jimallman]

My opinion after reading through this is that the Phylogenetic Justification actually fits best as a text field that is entered in association with the Calibration, rather than the Fossil Specimen. Is this possible, and if so do people agree with that idea.

It's possible, but would take a bit of effort. I think it's a question of how (absent the database) you would present this information in a calibration that uses multiple fossils. Some options:

1. Preserve the data as-is, but change the presentation. We might elevate a single fossil's phylo-justification to its own heading on the page. So the simple cases stay simple.

   If multiple fossils are present, we could list their justifications under this heading, naming the associated fossil in each case. If it's important to choose a "favorite" fossil in these cases, we might need to add a field to the calibration table that lets you choose this favorite. (If there are 5 fossils, can I pick 2 favorites, ie, two that really matter? Or is there always just the one?)

2. Move the phylo-justification to the calibration record, as Dan suggests above. In this case, we'd lose the ability to explain per-fossil justifications, which might lose something if there are several fossils involved and just one or two "favorites" described in the summary.
3. Add a new summary(?) justification to the calibration record. This would preserve the per-fossil justification if these are important, while giving the submitter a chance to summarize and prioritize their assessment, add caveats, etc.

[pdpolly]

Hi Jim and Dan,

Storagewise, it sounds like the existing database is compatible with Dan's request, as well as the other desirable features. Fossil only needs to be entered once, but it could have many links to calibrations through different phylogenetic justifications.

I do not have a strong opinion about where the entry fields for the justification are placed -- on the calibration page seems logical. It seems that the author's favored calibration will always hinge on one fossil (or at least on the fossils from one locality), no matter how many the author thinks are relevant for discussion. Therefore I prefer Option 1. We could display only the primary fossil on ordinary pages, retrieving the secondary fossils only in downloads of data.

pdp

On 9 Oct 2014, at 10:49 AM, Jim Allman notifications@github.com wrote:

My opinion after reading through this is that the Phylogenetic Justification actually fits best as a text field that is entered in association with the Calibration, rather than the Fossil Specimen. Is this possible, and if so do people agree with that idea.

It's possible, but would take a bit of effort. I think it's a question of how (absent the database) you would present this information in a calibration that uses multiple fossils. Some options:

Preserve the data as-is, but change the presentation. We might elevate a single fossil's phylo-justification to its own heading on the page. So the simple cases stay simple.

If multiple fossils are present, we could list their justifications under this heading, naming the associated fossil in each case. If it's important to choose a "favorite" fossil in these cases, we might need to add a field to the calibration table that lets you choose this favorite. (If there are 5 fossils, can I pick 2 favorites, ie, two that really matter? Or is there always just the one?)

Move the phylo-justification to the calibration record, as Dan suggests above. In this case, we'd lose the ability to explain per-fossil justifications, which might lose something if there are several fossils involved and just one or two "favorites" described in the summary.

Add a new summary(?) justification to the calibration record. This would preserve the per-fossil justification if these are important, while giving the submitter a chance to summarize and prioritize their assessment, add caveats, etc.

— Reply to this email directly or view it on GitHub.

[jimallman]

NOTE that some of @pdpolly's latest comment above is being hidden, and I'm not sure why. Click the ellipsis to see the paragraph that begins "I do not have a strong opinion..."

[jimallman]

If it's important to choose a "favorite" fossil in these cases, we might need to add a field to the calibration table that lets you choose this favorite. (If there are 5 fossils, can I pick 2 favorites, ie, two that really matter? Or is there always just the one?)

@pdpolly, what are your thoughts on this need to mark one or more "favorite" fossils for the current calibration? (I'm sure there's a better word for this, suggestions are welcome.) If one favorite is sufficient, an easy solution would be to feature the phylo-justification of the first fossil entered, and treat the others as secondary. This would of course be made clear on the data-entry screen.

[pdpolly]

I defer to Dan. The additional fossils will be of interest primarily to those doing analytical work on the confidence of the calibration. Otherwise the philosophy of the FCD has been "keep it simple", which is probably an argument for only displaying the justification for the single fossil that establishes the minimum age of the node.

On 10 Oct 2014, at 9:14 AM, Jim Allman notifications@github.com wrote:

If it's important to choose a "favorite" fossil in these cases, we might need to add a field to the calibration table that lets you choose this favorite. (If there are 5 fossils, can I pick 2 favorites, ie, two that really matter? Or is there always just the one?)

@pdpolly, what are your thoughts on this need to mark one or more "favorite" fossils for the current calibration? (I'm sure there's a better word for this, suggestions are welcome.) If one favorite is sufficient, an easy solution would be to feature the phylo-justification of the first fossil entered, and treat the others as secondary. This would of course be made clear on the data-entry screen.

— Reply to this email directly or view it on GitHub.

[Ksepka]

Hi all - I'm heading to the airport for the Missouri Botanical Garden Symposium ( will advertise FCD). In brief I think the first entered fossil should be "Primary Fossil" and the only to display under the Keep it Simple philosophy. I'm ok with "Add Additional Fossil" in the data entry for those that want to use it.

Sent from my iPhone

On Oct 10, 2014, at 10:01 AM, "P. David Polly" notifications@github.com wrote:

I defer to Dan. The additional fossils will be of interest primarily to those doing analytical work on the confidence of the calibration. Otherwise the philosophy of the FCD has been "keep it simple", which is probably an argument for only displaying the justification for the single fossil that establishes the minimum age of the node.

On 10 Oct 2014, at 9:14 AM, Jim Allman notifications@github.com wrote:

If it's important to choose a "favorite" fossil in these cases, we might need to add a field to the calibration table that lets you choose this favorite. (If there are 5 fossils, can I pick 2 favorites, ie, two that really matter? Or is there always just the one?)

@pdpolly, what are your thoughts on this need to mark one or more "favorite" fossils for the current calibration? (I'm sure there's a better word for this, suggestions are welcome.) If one favorite is sufficient, an easy solution would be to feature the phylo-justification of the first fossil entered, and treat the others as secondary. This would of course be made clear on the data-entry screen.

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[jimallman]

In brief I think the first entered fossil should be "Primary Fossil" and the only to display under the Keep it Simple philosophy.

Love it! For now, I'll hide any other fossil justifications in the calibration display page, but other information about these fossils will still appear under "fossils used to date this node". If you prefer, I can hide them here as well.

[pdpolly]

Sounds good. But I suggest that there be a specific toggle for "primary fossil". By default it could be auto-ticked for first entry, but it is easy to imagine a situation where one needs to change which one is designated as primary (e.g., editor entered them in the wrong order, author revises calibration, redating results in new primary fossil, etc.)

On 10 Oct 2014, at 11:33 AM, Jim Allman notifications@github.com wrote:

In brief I think the first entered fossil should be "Primary Fossil" and the only to display under the Keep it Simple philosophy.

Love it! For now, I'll hide any other fossil justifications in the calibration display page, but other information about these fossils will still appear under "fossils used to date this node". If you prefer, I can hide them here as well.

— Reply to this email directly or view it on GitHub.

[jimallman]

So just one primary fossil, never more than one?

(I'll treat a single fossil as primary, in any case.)

[pdpolly]

yes, only one primary fossil. logically speaking, there only needs to be one to establish the best estimate of the minimum age of a node.

On 10 Oct 2014, at 1:21 PM, Jim Allman notifications@github.com wrote:

So just one primary fossil, never more than one?

(I'll treat a single fossil as primary, in any case.)

— Reply to this email directly or view it on GitHub.