[Taxacom] Molecular evidence (in 2017) supports my 2014 theory about rafting killifish

Kenneth Kinman kinman at hotmail.com
Sun Dec 27 14:56:54 CST 2020

Thanks Jason,
        Your post was very good.  It was then followed by Heads who (like Grehan) just repeats the same old arguments.  It is interesting that Heads quotes De Baets *et al*., 2016 as saying ‘could be considered a dark art’, but their source for that "dark art" comment was Heads himself (Heads, 2012; "Bayesian transmogrifcation...").
        I don't understand why Heads and Grehan have to bash those who disagree with them, using phrases like "dark art", twaddle, transmogrification, "geological data manipulation", "magical black box", and on and on.  Especially when it appears in scientific papers.
       By the way, I was reading a published reply to Heads, 2011 (in which he repeatedly accused his opponents of transmogrification).  In that reply, they say: "We demonstrate here that calibration using tectonic events, as suggested by Heads (2011), generates an unrealistic hypothesis where the evolution of Asteraceae must have taken place contemporary with or earlier than the Cambrian explosion. This would be hundreds of million years earlier than the origin of the asterids."
      This reply to Heads can be read here:  https://academic.oup.com/sysbio/article/61/3/522/1670654

From: Taxacom <taxacom-bounces at mailman.nhm.ku.edu> on behalf of Michael Heads via Taxacom <taxacom at mailman.nhm.ku.edu>
Sent: Sunday, December 27, 2020 1:26 PM
To: JF Mate <aphodiinaemate at gmail.com>
Cc: Taxacom <taxacom at mailman.nhm.ku.edu>
Subject: Re: [Taxacom] Molecular evidence (in 2017) supports my 2014 theory about rafting killifish

'...not a magical black box'.

No, it's definitely a magical black box. You can get any date you want.
Here's a section from a recent manuscript:

*The problem of the priors*

In the Modern Synthesis approach, ‘Fossils record the origins and
disappearance of organisms’ (Herrera, 2017). However, a clade’s actual age
is older than its oldest fossil’s age by an unknown amount of time. In most
studies, clade ages are calculated from fossil-calibrated molecular-clock
analyses. However, fossil calibrated studies on their own can provide only
minimum ages. In Bayesian analyses, these *fossil* ages are converted into
estimates of absolute (not minimum) *clade* ages by imposing priors that
stipulate how much older than its oldest fossil a clade is. Naturally, the
estimates of clade age are extremely sensitive to the time priors (Warnock *et
al*., 2015).

The problem is that the choice of the priors is left up to the author, and
the priors that are chosen are invariably very narrow; in other words, a
clade used for calibration is assumed to be not much older (often <10 Myr) than
its oldest fossil. This use of very narrow priors represents the formal
imposition of an old belief – that fossil age more or less equals clade age.

Choosing priors is ‘highly arbitrary’ (Pirie and Doyle, 2012), ‘...often
subjective…’ (Ho & Duchêne, 2014), ‘invariably established without
justification’ (Warnock *et al*., 2015) and, in sum, ‘could be considered a
dark art’ (De Baets *et al*., 2016). Thus, while the clade ages generated
in Bayesian studies are extremely precise and have excellent statistical
support, becaue of the problem of the priors they can be ‘grossly
inaccurate’ (dos Reis *et al*., 2016) and could be tens or even hundreds of
millions of years too young. As Hauenschild *et al.* (2018) observed, clade
ages that are too young may be calculated ‘based on too narrowly set
priors, and vicariance thereby may be erroneously ruled out’.

On Mon, Dec 28, 2020 at 2:59 AM JF Mate via Taxacom <
taxacom at mailman.nhm.ku.edu> wrote:

> John, you are not discussing the points nor answering the questions but
> simply repeating "twaddle" in an attempt to rescue a point of view
> (Croizatian Panbiogeography) which is simply untenable nowadays. All
> biogeographers consider vicariance and dispersal, and their definition of
> dispersal does not depend on accepting Panbiogeography.
> As for node calibration, exponential and lognormal bounds are not maximum
> ages but probability curves which extend the "minimum" node age into the
> past as a first step. The model then works the best fit based on the data
> at hand and gives you estimated ages with ranges. That is a transparent
> model, not a magical black box as you are claiming.
> It is fair to argue if the model parameters are adequate, if the fossils
> used are properly interpreted, or to reanalyze the data. These are all
> limitations, but simply using them to disregard studies out of hand because
> they don't support your point of view is not helpful.
> Best
> Jason
> On Sat, 26 Dec 2020, 01:57 John Grehan, <calabar.john at gmail.com> wrote:
> >
> > Jason – thank you for elaborating on the defense of minimums
> > misrepresented as maximums. Comment below:
> >
> > “0.5 SD on the fossil's age seems a reasonable and conservative range”
> >
> > 'reasonable or not' it's still just made up out of thin air. It is magic.
> >
> > “Are you arguing for a minimum hardbound and if so what is your
> > justification? “
> >
> > Yes, although not sure what you mean by 'hardbound'.  A minimum is a
> > minimum so it is hardbound in that respect. Justification is simple,
> > fossils provide minimum ages and no data on potential older ages of
> origin
> > for the taxon in question.
> >
> > “why not use lognormal or exponential for both points and maybe some
> > sensitivity analysis in regards to this but other than that I wouldn't
> call
> > it twaddle”
> >
> > You are welcome to use anything you like, but it's still magic and the
> > result is twaddle. A minimum cannot be turned into a maximum. That is the
> > reality. The rest is an invention of the imagination.
> >
> > “This argument is at best specious. Molecular calibrations have been
> using
> > soft calibrations for well over a decade now to account for the
> > uncertainties that you mention “
> >
> > Wrong, the molecular calibrations misrepresent minimum ages as maximum.
> >
> > “ majority use either lognormal or exponential bounds “
> >
> > So what? They are both inventions of the mind – magical transformation of
> > minimum into maximum.
> >
> > “In other words, there is already the acknowledgement that known fossils
> > may not be the oldest exemplars of a taxon, but you have to work with
> what
> > you have
> >
> > of course. But if fossils may not be the oldest exemplars then the
> > molecular extrapolations may not be either. To say otherwise is to
> invoke a
> > magical transformation - not science as I understand it.
> >
> > “ and, when new fossil evidence is discovered, corrections will be made.”
> >
> > Corrections to what? Corrections to a maximum that never was in the first
> > place?
> >
> > “What you are asking for is the legal equivalent of proving someone's
> > innocence.”
> >
> > Rubbish. I am asking that the empirical reality be recognized – that
> > fossil calibrated molecular divergence dates be recognized as minions
> that
> > cannot falsify earlier origins or biogeographic correlations that may
> > suggest earlier origins.
> >
> > Molecular transformations of minimums into maximums is perhaps the only
> > modern case of magic being used in science that I am aware of (other than
> > perhaps some political inventions with COVID). Just to be sure that there
> > is no misunderstanding (as has happened in the literature),
> > panbiogeographic approaches do not ignore or disregard molecular
> divergence
> > ages, but accept and incorporate those findings as MINIMUM ages only.
> That
> > is science.
> >
> > Cheers,
> >
> > John Grehan
> >
> >
> > On Fri, Dec 25, 2020 at 7:01 AM JF Mate <aphodiinaemate at gmail.com>
> wrote:
> >
> >> Michael, regarding the calibration method used: "It corresponds to the
> >> origin of the crown European cyprinodontoid clade [39], which was
> estimated
> >> to have occurred at least 33 Ma on the basis of the oldest identifiable
> >> clade member, the
> >> fossil Prolebias stenoura Sauvage, 1874 from the Lower Stampien (Lower
> >> Oligocene) of Puy-
> >> de-Dome, France [40] (prior setting: lognormal distribution, mean = 33
> >> and standard devia-
> >> tion = 0.5). ( [40] Costa WJEM. Oligocene killifishes (Teleostei:
> >> Cyprinodontiformes) from southern France: relationships, taxonomic
> >> position, and evidence of internal fertilization. Vertebr Zool. 2012;
> 62:
> >> 371–386.)
> >>
> >> John:
> >> 0.5 SD on the fossil's age seems a reasonable and conservative range,
> >> with a lognormal distribution for the node. My only quibble is why not
> use
> >> lognormal or exponential for both points and maybe some sentivity
> analysis
> >> in regards to this but other than that I wouldn't call it twaddle. Are
> you
> >> arguing for a minimum hardbound and if so what is your justification?
> >>

More information about the Taxacom mailing list