[Taxacom] Taxacom Digest, Vol 185, Issue 14

Mon Sep 20 08:05:10 CDT 2021

Hello John and Michael!

Below are my thoughts. Your text is set in italics to differentiate from my
responses.

Cheers!
Brendon

*Hi Brendon,  *

*Great to have your comment and viewpoint. Taken in a non-combative spirit
as always. When you say you disagree, I presume you are disagreeing with my
assertion that priors are just guesses. At this point your description does
not help me see anything to the contrary. You say that you "set my fossil
priors as the evidence-based stratigraphic range for the given deposit or
formation." So what does that really mean? *

That means that I do not enter an arbitrary guess for the age of a fossil.
For example, if I were to conduct a tip-dating (“total-evidence”)
phylogenetic analysis of the Dolichoderinae, I would want to include †*Azteca
alpha*, from Dominican ambers. I do not enter “Azteca_alpha (100, 90)” or
“Azteca_alpha (65, 20)” or “Azteca_alpha (65, 0)”, as these would indeed be
guesses. Rather, I consider the set of studies which have evaluated the
possible age range of the Dominican formations and choose the age range
which is best supported. I would not set “Azteca_alpha (45, 40)”, the NMRS
estimate of Lambert, Frye, and Poinar (1985, *Archaeometry*, 27, 43),
because this study employed only two points to for their age calibration
curve. I would choose instead “Azteca_alpha (20, 15)”, based on the
multiple lines of corroborating evidence provided by Itteralde-Vinent and
MacPhee (1996, *Science*, 273, 1850–1852). If I were concerned about the
conflicting ages, I could enter “Azteca_alpha (45,15)” and evaluate the
possible effects this has on the posterior distribution of trees. A final
alternative is to just use “Azteca_alpha (15)” to say that the minimum age
for the fossil is 15 million years, although I would prefer to use date
ranges.

*I actually have no problem with anyone assigning priors on how much older
than the oldest fossil they think a taxon might be, or even making up
probabilities for that. The problem arises when these are transformed into
empirical limits that are supposed to falsify earlier origins predicted
from tectonic correlation (for example). Also, the means of the priors are
very often presented as actual ages so that taxa are said to have
originated at (or about) such and such a date (and precluding earlier
tectonically mediated origins). *

When used as node calibrations, the lognormal and normal distributions
allow the analysis to estimate ages which are effectively infinitely old
for that given node. If the data are informative, the posterior estimate
for the node’s age will narrow, thus more incisive for hypothesis testing;
if the data are uninformative, the posterior range will be broad. I would
not use a prior to falsify a hypothesis; such a construction is an invalid
hypothesis test. The test can be done through the comparison of model fit
or log likelihood across multiple analyses, with the caveat that there are
no final answers. The hypothesis testing is done with the posterior
estimates, not the priors.

*I had one colleague note to me that one of the probability models (might
be long normal, not sure) did not set an upper limit on the possibilities,
yet in practice that is what most authors do when they say they refute
earlier origins predicted by tectonic correlation. *

What is it that most authors do when they claim refutation of tectonic
correlation? I am not sure I follow this thought.

*I might have less of an issue if priors were presented in the form of a
probability that the oldest age, or even the mean age, was AT LEAST such
and such age. *

Priors set for fossil ages in the tip-dating framework are effectively
statements of “this fossil is approximately X years old”. The priors of
fossil ages are just one set of parameters in the analysis, which are then
used to estimate the branch lengths given the available evidence. Once a
Bayesian analysis is run, one receives a range of posterior probabilities,
for which the only statistically honest way to present them are as ranges.
The interpretation from there can vary and should be justified.

*One might take the view that priors make Mesozoic origins for extant taxa,
even species or genera, very ‘improbable’, but that is not the same as
impossible, and heck, there are an awful lot of Mesozoic (not to mention
Cenozoic) tectonic correlations. *

Setting a prior on a node itself based on the minimum age of a fossil of
presumed relationship is “node calibration”, and can have several
distributions, most of which have infinite tails. I am not sure about the
statement that many extant species have a Mesozoic origin. There is no
direct or indirect evidence, for example, that socially parasitic ant
species which render their host species paraphyletic are > 65 million years
old.

*To attribute that all to ‘chance’ coincidence (which would seem to be the
only explanation) is hardly productive. It’s kind of like saying that God
only made all the evidence of Creation only look like evolution occurred.*

I am not sure about this statement. Stochastic events can have critical
effects on populations or sets of populations.

*Cheers, John*

*Hi Brendon,*

*You say 'I set my fossil priors as the evidence-based stratigraphic range
for the given deposit or formation'.*

*That is, if the fossil is in a formation dated as extending from 10-20 Ma,
the group can be no older than 20 Ma. But why not? I don't see that the age
of the formation is relevant. It's like saying that if a fossil is in the
Cenozoic, it's group can't be Mesozoic, because that's a separate era. .*

I did not state that a group cannot be older than the oldest estimated age
for a given fossil deposit. The age of a fossil’s formation is relevant
when one is conducting tip-dating (“combined-evidence”) analysis, which
incorporates the ages of the given fossils themselves in the estimation of
the age distributions of the tree. In these analyses, the placement of the
fossil is allowed to vary and is informed by the morphological data. One
can choose to use node calibrations in tip-dating analysis or not. In
tip-dating analysis

*You're right, this technique ('bounding') isn't guesswork, but it seems
arbitrary and illogical. Why not just say that, based on the fossil record,
the fossil's group is older than the fossil by some unknown amount? This is
a valid and useful contribution, and data from other fields (e.g.
biogeography) can then be used to investigate further. I don't think you
can assess the accuracy of the fossil record by using data from the same
record.  *

One should indeed expect that the age of a fossil group is older than the
oldest crown fossil of that group. It is precisely the age information of
the fossil which is used to estimate a range of dates for nodes of the
tree. The question of fossil record accuracy is distinct from the use of
the phenotypic and stratigraphic data from a sample of fossils for
phylogenetic analysis.

*Biogeographic-tectonic calibration  is now used quite widely (e.g. Ho et
al., 2015; De Baets et al., 2016; Landis, 2017; Gunter et al., 2018, Pett &
Heath, 2020).  As Landis (2020) argued, ‘Fossil and biogeographic dating
methods… should be regarded as complementary, not competing, strategies’.
This is the synthetic approach we take, using fossil-calibrated ages as
estimates of minimum clade age and biogeographic-tectonic calibrations as
estimates of actual clade age.  *

I have no comment for the use of tectonic splits in biogeographic analysis,
except to note that I do not think that plate tectonics influences the
evolution of socially parasitic ants of the Colorado plateau.

On Sun, Sep 19, 2021 at 9:51 PM Michael Heads <m.j.heads at gmail.com> wrote:

> Hi Brendon,
>
> You say 'I set my fossil priors as the evidence-based stratigraphic range
> for the given deposit or formation'.
>
> That is, if the fossil is in a formation dated as extending from 10-20 Ma,
> the group can be no older than 20 Ma. But why not? I don't see that the age
> of the formation is relevant. It's like saying that if a fossil is in the
> Cenozoic, it's group can't be Mesozoic, because that's a separate era. .
>
> You're right, this technique ('bounding') isn't guesswork, but it seems
> arbitrary and illogical. Why not just say that, based on the fossil record,
> the fossil's group is *older than* the fossil by some unknown amount?
> This is a valid and useful contribution, and data from other fields (e.g.
> biogeography) can then be used to investigate further. I don't think you
> can assess the accuracy of the fossil record by using data from the same
> record.
>
> Biogeographic-tectonic calibration  is now used quite widely (e.g. Ho et
> al., 2015; De Baets et al., 2016; Landis, 2017; Gunter et al., 2018, Pett &
> Heath, 2020).  As Landis (2020) argued, ‘Fossil and biogeographic dating
> methods… should be regarded as complementary, not competing, strategies’.
> This is the synthetic approach we take, using fossil-calibrated ages as
> estimates of minimum clade age and biogeographic-tectonic calibrations as
> estimates of actual clade age.
>
> On Mon, Sep 20, 2021 at 3:52 AM Brendon E. Boudinot via Taxacom <
> taxacom at mailman.nhm.ku.edu> wrote:
>
>> Hello John!
>>
>> I disagree. In tip-dating analyses, I set my fossil priors as the
>> evidence-based stratigraphic range for the given deposit or formation.
>> Using this range, we can account for patterns of variation for both
>> morphology and molecular data, given the estimated range of possible
>> relationships and clades ages. The priors have strong affect when there is
>> little information, such as at a root node. The main point is accounting
>> for uncertainty regarding the most recent or oldest possible age for a
>> fossil. Please explain, in this context, why that is a problem. I am here
>> today because of curiosity in general; likewise, I am responding now
>> because of specific curiosity. What alternatives should one make use of to
>> combine genotypic, phenotypic, and stratigraphic ranges for estimating a
>> possible set of relationships and ages?
>>
>> Noncombatively,
>> Brendon
>>
>> On Sat, 18 Sep 2021, 19:05 <taxacom-request at mailman.nhm.ku.edu> wrote:
>>
>> > Daily News from the Taxacom Mailing List
>> >
>> > When responding to a message, please do not copy the entire digest into
>> > your reply.
>> > ____________________________________
>> >
>> >
>> > Today's Topics:
>> >
>> >    1. Re: Snake garbage (John Grehan)
>> >
>> >
>> > ----------------------------------------------------------------------
>> >
>> > Message: 1
>> > Date: Fri, 17 Sep 2021 16:25:36 -0400
>> > From: John Grehan <calabar.john at gmail.com>
>> > To: taxacom <taxacom at mailman.nhm.ku.edu>
>> > Subject: Re: [Taxacom] Snake garbage
>> > Message-ID:
>> >         <CADN0ud1-28JXFunp23V6=
>> > v4evN7orEypWnxjf04t+TGoES5bEQ at mail.gmail.com>
>> > Content-Type: text/plain; charset="UTF-8"
>> >
>> > I wrote the original note in a bit of a hurry (should not do that) and
>> > quite justifiably got castigated off list for not making sense. So
>> below a
>> > hopefully more coherent rant.
>> >
>> > This paper carries on the time honored scientific tradition of just
>> > ignoring shortcomings of a method and ploughing on as if the ground was
>> not
>> > already falling away beneath. In this case the paper does this by
>> > representing priors as some kind of empirically real source of
>> estimating
>> > fossil calibrated clade ages whereas it has been shown that fossil
>> > calibrated ages cannot be anything but a minimum ages. Priors are just
>> > personal guesses about how much older than the older fossil a taxon
>> might
>> > be (dressed up in numbers to look scientific). These authors just ignore
>> > that. And they continue the temptation of using an automated
>> biogeography
>> > program as 'evidence' despite its inherent inability to distinguish
>> between
>> > vicariance and dispersal where either can generate the same
>> biogeographic
>> > pattern. In other words, they use a plug and play program that can
>> render
>> > artificial results and they have no way to know. But the authors carry
>> on
>> > the pretense that this is not the case.
>> >
>> > Cheers, John
>> >
>> > On Fri, Sep 17, 2021 at 10:20 AM John Grehan <calabar.john at gmail.com>
>> > wrote:
>> >
>> > > "Evolution and dispersal of snakes across the Cretaceous-Paleogene
>> mass
>> > > extinction" (https://www.nature.com/articles/s41467-021-25136-y.pdf )
>> > >
>> > > This paper carries on the time honored scientific tradition of just
>> > > ignoring shortcomings of a method and ploughing on as if the ground
>> was
>> > not
>> > > already falling away beneath. In this case the representation or
>> priors
>> > as
>> > > some kind of empirically real source of estimating fossil calibrated
>> > clade
>> > > ages as anything but a minimum ages, and the continued temptation of
>> > using
>> > > an automated biogeography program as 'evidence' despite its
>> > > inherent inability to distinguish between vicariance and dispersal
>> where
>> > > either can generate the same biogeographic pattern. I have been
>> attacked
>> > > for calling this stuff 'garbage' but I have not come up with a more
>> > > accurate term - yet.
>> > >
>> > > John Grehan
>> > >
>> >
>> >
>> > ------------------------------
>> >
>> > Subject: Digest Footer
>> >
>> > Taxacom Mailing List
>> >
>> > Send Taxacom mailing list submissions to taxacom at mailman.nhm.ku.edu
>> > For list information; to subscribe or unsubscribe, visit:
>> > http://mailman.nhm.ku.edu/cgi-bin/mailman/listinfo/taxacom
>> > You can reach the person managing the list at:
>> > taxacom-owner at mailman.nhm.ku.edu
>> > The Taxacom email archive back to 1992 can be searched at:
>> > http://taxacom.markmail.org
>> >
>> > Nurturing nuance while assailing ambiguity for about 34 years,
>> 1987-2021.
>> >
>> >
>> > ------------------------------
>> >
>> > End of Taxacom Digest, Vol 185, Issue 14
>> > ****************************************
>> >
>> _______________________________________________
>> Taxacom Mailing List
>>
>> Send Taxacom mailing list submissions to: taxacom at mailman.nhm.ku.edu
>> For list information; to subscribe or unsubscribe, visit:
>> http://mailman.nhm.ku.edu/cgi-bin/mailman/listinfo/taxacom
>> You can reach the person managing the list at:
>> taxacom-owner at mailman.nhm.ku.edu
>> The Taxacom email archive back to 1992 can be searched at:
>> http://taxacom.markmail.org
>>
>> Nurturing nuance while assailing ambiguity for about 34 years, 1987-2021.
>>
>
>
> --
> Dunedin, New Zealand.
>
> My books:
>
> *Biogeography and evolution in New Zealand. *Taylor and Francis/CRC, Boca
> Raton FL. 2017.
> https://www.routledge.com/Biogeography-and-Evolution-in-New-Zealand/Heads/p/book/9781498751872
>
>
> *Biogeography of Australasia:  A molecular analysis*. Cambridge
> University Press, Cambridge. 2014. www.cambridge.org/9781107041028
>
>
> *Molecular panbiogeography of the tropics. *University of California
> Press, Berkeley. 2012. www.ucpress.edu/book.php?isbn=9780520271968
>
>
> *Panbiogeography: Tracking the history of life*. Oxford University Press,
> New York. 1999. (With R. Craw and J. Grehan).
> http://books.google.co.nz/books?id=Bm0_QQ3Z6GUC
> <http://books.google.co.nz/books?id=Bm0_QQ3Z6GUC&dq=panbiogeography&source=gbs_navlinks_s>
>
>
>
>
>
>
>
>
>
>
>

-- 
Dr. Brendon E. Boudinot
Alexander von Humboldt Research Fellow
Friedrich-Schiller-Universität Jena
Institut für Zoologie und Evolutionsforschung
Erbertstraße 1
07443 Jena DE