From:	Robert Hazen <rhazen@ciw.edu>
Sent:	Friday, March 25, 2016 15:11
To:	Dimitri Sverjensky; Jihua Hao; Paul Falkowski; Andrew Knoll; Peter 
Fox; Michael Meyer; Mark Ghiorso; Elisha K. Moore; Sophie 
Kolankowski; Marshall X Ma
Subject:	Fwd: Appalachian State notes
Attachments:	num ima minerals-vs-num elements -mindat-raw-20160115.docx

Flag Status:	Flagged

Dear Colleagues,

I wanted to key you into an email conversation related to mining big mineral data resources. 
Some of us have been thinking about reasons for observed mineral diversity. One obvious factor 
is the geochemical complexity of a locality. The following discussion and attach graph are self-
explanatory.

Any thoughts?

Bob



---------- Forwarded message ---------- 
From: Robert Hazen <rhazen@ciw.edu> 
Date: Fri, Mar 25, 2016 at 3:06 PM 
Subject: Re: Appalachian State notes 
To: Daniel Hummer <dhummer@carnegiescience.edu> 
Cc: "Golden, Joshua J - (jgolden)" <jgolden@email.arizona.edu>, "Downs, Robert T - (rdowns)" 
<rdowns@email.arizona.edu>, Grethe Hystad <Grethe.Hystad@purduecal.edu>, Edward Grew 
<esgrew@maine.edu>, Chao Liu <cliu@carnegiescience.edu>, "Morrison, Shaunna - 
(shaunnamm)" <shaunnamm@email.arizona.edu> 

Hi All,

First, thank you Joshua! You have become a real master in mining these data resources.

Yes, there is definitely something here of interest--an insight into origins of mineral diversity as 
a function of element diversity.  Just to amplify a bit, there are about 12 ur-minerals using 10 
essential elements; 60 chondrite minerals using 20 elements; 250 meteorite minerals with (I think) 
about 35 elements; and today 5100 minerals with 72 elements. The data Josh compiled on 
individual localities seem to have a similar trend.

Can we say something similar about the Moon and Mars? Could this be a way to predict how 
many minerals one might find on another world--essentially a function of geochemical 
differentiation?

And regarding individual mineral localities, at the lower end of species diversity (left-hand limits) 
are a bunch of places in mindat where there are only a few obvious minerals reported. These 
localities really are incomplete and can't tell us much. 

On the other hand, the right-hand limit seems to be telling us something significant, and the 
curve definitely shows the predicted nonlinear increase in numbers of species versus numbers of 
essential elements. There is some kind of combinatorial effect. So, can we come up with a 
rational way to identify the more thoroughly studied localities? And what is the mathematical 
form of the right-limit curve? And is there a theoretical reason for the curve? 

Will be thinking about these matters. Let me know what you think.

Best,

Bob


On Fri, Mar 25, 2016 at 2:44 PM, Daniel Hummer <dhummer@carnegiescience.edu> wrote:
That graph IS really interesting.... there seems to be a very well defined upper and lower 
envelope. I wonder if there's a theoretical or statistical reason for that. 
 
Dan 
 

Daniel R Hummer
Postdoctoral Scholar, Mineralogy and Crystallography
Geophysical Laboratory
Carnegie Institution of Washington
Phone: (814) 321-8859
Fax: (202) 478-8901
On 3/25/16 2:21 PM, Golden, Joshua J - (jgolden) wrote:
Dear all,
I have attached a scatter plot of number of minerals vs number of elements from 
the mindat dataset of 137824 localities. This plot is from the original mindat 
export file from 20160115. This number of localities differs slightly than what is 
currently on the database (135769) due to the fact that on the database we have 
removed "duplicate" mineral entries from localities that contain a mineral at both 
the bottom level and one of the parents.
I know this is not exactly what we want, but is the first order of the data.
The graph looks very interesting.
Best regards,
Josh
 
From: Robert Hazen <rhazen@ciw.edu> 
Sent: Thursday, March 24, 2016 6:23 PM 
To: Downs, Robert T - (rdowns); Golden, Joshua J - (jgolden); Grethe Hystad; Edward 
Grew; Daniel Hummer; Chao Liu; Morrison, Shaunna - (shaunnamm) 
Subject: Appalachian State notes 
 
Dear Colleagues,

I'm just back from Appalachian State, where I gave 2 seminars, 1 public lecture, and 
spent many hours meeting with students and faculty, mostly on mineral 
evolution/ecology themes. Very nice reception from faculty and students; also 
lots of enthusiasm for changing the way mineralogy is taught. 

Two things of interest to our team. First, their small geology teaching 
museum recently received a nice donation of ~200 fine mineral specimens. 
The new displays were unveiled as part of my visit and were organized 
around mineral evolution (photos attached). The displays are small but 
very effective and I was quite touched at their efforts and the little 
ceremony to dedicate it.

Second, after my mineral ecology seminar (remarkably well-attended geology event, with 
150 people jammed into a room that seated ~130), an ecologist (Bob Creed) came up 
and discussed aspects of island biogeography in more detail. Turns out it’s not a strictly 
log-log relationship between number of species and island area, because for larger 
islands you get coupling effects that lead to relatively more species. While I don’t see a 
direct correlation to mineral diversity, it did trigger an idea.
 
What is the relationship between the number of mineral-forming elements (N) at a locality 
and the number of observed species (S) at that locality (assuming by locality we mean a 
well-defined mineral-forming district)? In a strictly equilibrium world the answer is Gibbs’ 
phase rule: S ? N + 2. Thus, a granite with Si, Al, Na, K, Fe, Mg, Ca, Ti, Zr, H, and F 
typically will have ten different minerals [e.g., quartz, 3 feldspars (including perthitic 
exsolution), 2 micas, magnetite, sphene, zircon, fluorite]. 
 
But many localities have multiple pulses of mineralization, secondary alteration and 
weathering, and metastable persistence of phases. So I’m guessing that if we plot N 
versus S for well-researched localities (though I’m not sure how one determines which 
mineralized districts are “well-researched”), we’ll find a distinctly non-linear trend, with far 
more minerals occurring at localities with lots of mineral-forming elements. This idea 
follows from the combinatorial richness of chemistry and the metastability of many 
mineral phases. 
 
I propose that we start such an effort by tabulating those localities/districts in mindat that 
have the greatest numbers of essential elements and then list the numbers of known 
species from those localities. My guess is that the top 1000 localities in terms of numbers 
of essential elements will also tend to be well researched in terms of their minerals. I 
realize that mindat.org isn't flawless in its lists of minerals, but perhaps we can target 
those localities that have reliable monographic treatment of their mineralogies. Then we 
can see if there’s any trend at all by plotting N versus S for those localities/districts.
 
This idea isn’t a high priority, but it would be an interesting additional factor in trying to 
understand diversity-distribution relationships. If there's a relatively easy way to mine 
mindat then I'd suggest we do a trial plot and see if there's an interesting trend.

Be well; hope to see most of you in the next week.

Bob
 


-- 
Robert M. Hazen
Senior Staff Scientist, Geophysical Laboratory
Executive Director, Deep Carbon Observatory
5251 Broad Branch Road NW
Washington, DC 20015
phone: 202-478-8962
e-mail: rhazen@ciw.edu

Personal web site: http://hazen.gl.ciw.edu
DCO website: deepcarbon.net 
Keck Deep-Time Project website: http://dtdi.carnegiescience.edu