Water analysis 'not detected'

[simonh]

Hi Ron,

First let me thank you for all the work put into this study and enormous amount of time it must have taken compiling your first article.

On your analysis I noticed some of the elements measured were 'Not detected'. I presume this means that they were below the detection limits of the analysis.

Do you have the detection limits for those analyses? I was interested in Iron for example. On these is the detection limit below the NSW concentration (i.e. the 'true value' must therefore be below the NSW levels, so we could classify them as 'less than' relative NSW) or is the detection limit above the NSW levels (i.e we classify these element as 'not known' relative to NSW)?

Hope my question makes sense.

[Randy Holmes-Farley]

Simon:

The detection limits should absolutely be given. This is most critical for elements where the detection limit is actually HIGHER than surface NSW levels (like iron). For these ions, the tank levels may still be above NSW levels, but be below detectable limits.

[cvye]

Excellent study!
However, I would have added an additional control group (given that the majority of subjects used Instant Ocean as a salt mix):

IOSW - say a vat of Instant Ocean based seawater, perhaps aerated for 48 hrs and re-adjusted for salinity using RO/DI.

Be that as it may, the conclusion that our tanks are not even close to natural seawater is an old one, I believe, credited to Steve Spotte?

What will be much more interesting will by your future article(s) describing the tanks with "problems" and those without and which ones those were in the spectrum! Can't wait!

[rshimek]

Hi,

All of the detection limits will be given in the next article.

For iron the D. L. was 0.005 mg/l

My table lists NSW sea water as 0.02 to 0.002 mg/l

cvye,

The IO values in the study DO come from a sample of IO made up as described in the materials and methods.

:D

[cvye]

oops, dopey me, I missed that which is so clear on re-reading. :o

[tatuvaaj]

Dr. Shimek,

First let me thank you for your great work! These types of study can really benefit hobbyists!

Comments:

Your data for NSW levels seem to be quite different from tables I have (for example see this excellent page with references: http://www.dnai.com/~patwilde/ocpertbl.html)

Was your data compiled from reef waters or was there any other reason to select that study as a reference?

The data from 1966 seems quite old. In the book "An introduction to the chemistry of the sea" by Michael E. Q. Pilson he states that no data published about trace elements before mid 70s should be trusted. The link I provided seems to have more recent data for example for Fe (4.50E-05 ppm).

Once again, thank you very much for the hard work you have done!

[Randy Holmes-Farley]

Surface water iron values (in the top 50 meters) are much lower than the water column overall due to sequestration by growing organisms. Values can be equal to 0.1 nM (0.000006 ppm).

[Randy Holmes-Farley]

Tatu:

Oops, I guess we posted the same thing at the same time.

[tatuvaaj]

Randy,

:)

Do you agree with Pilson about trace metal measurements before say 80s being unreliable?

[rshimek]

Hi Guys,

[thanks]

Tatu,

I will be glad to use the link you provided to get more up-to-date data for the subseqent articles about these data.

The data I used simply came from a reference on hand, and were originally from Svedrup's oceanography text.

Randy,

If you have any other suggestions for obtaining up-to-date data online, I could certainly use them.

:D

[tatuvaaj]

Even the data in that web page is old when compared to the Pilson's book I mentioned earlier - he has compiled data for trace elements from several studies mainly from the 90s. So if you happen to have access to that book it might be a good source.

I just compared the data in Pilson's book to the "An introduction to marine biochemistry" by Susan M. Liebes (as you can see, I'm at the "An introduction..." level :) ) and they seemed to be reasonable close to each other. Liebes' data comes from "Chemical Oceanography", vol. 8, K. W. Bruland 1983 pp. 172-173

[Randy Holmes-Farley]

Tatu:

Your comment about old studies could certainly be true. Both collection and analysis techniques are much better now.

One concern that I have about using any single number is that for many of these ions there is a profile that changes greatly with depth. Sometimes the difference is several orders of magnitude. I'm not sure how to deal with that in this context, except to try to compare to surface water values.

Ron:

Unfortunately, I don't know of any good online tables. Millero's book "Chemical Oceonography" (1996) is pretty good with tables and graphs of data.

[rshimek]

Hi Randy,

I was afraid you were going to say that.

Since my initial forays into chem-ocean classes thirty years ago, I have avoided purchasing new text... and I surely can't get my hands on a copy of one at the local University (know to one and all as, Moo -U (or mebbe moo-ewe).

I share your concerns with comparing to NSW, actually, and the thrust of this study is not that comparsion, but rather a comparison within reef tanks, and I hope to use natural data sparingly unless there are very great differences.

Additionally, surface mid-ocean waters are likely to be very different from reef waters, and reef data are damnably hard to come by.

:D

[tatuvaaj]

I don't know the rules about posting scientific data so I hope they don't lock me in the jail...

Edited:
Oops, I guess you people are not using UNIX so the attachment isn't very clear ;)
Here is the data:
<pre>
Mean Range Distribution
Li 25 umol - C
Be 20 pmol 4-30 N;Sc
N 30 umol 0.1-45 N
Al 10 nmol <0.1-40 Mm
Si 100 umol <1-200 N
P 2.3 umol <0.1-3.5 N
Sc 15 pmol 8-20 Sd
Ti 200 pmol 4-300 Sd
V 30 nmol 20-35 Sd
Cr 4 nmol 2-5 N
Mn 0.5 nmol 0.2-3 Dd
Fe 1 nmol 0.1-2.5 Sd;Dd
Co 20 pmol 10-100 Sd;Dd
Ni 8 nmol 2-12 N
Cu 4 nmol 0.5-6 N;Sc
Zn 6 nmol 0.05-9 N
Ga 20 pmol 2-50 Cx;Sc
Ge 70 pmol <7-115 N
As 23 nmol 15-25 N
Se 1.7 nmol 0.5-2.3 N
Rb 1.4 umol - C
Y 250 pmol 80-300 ?
Zr 200 pmol 12-300 N;Sc
Mo 100 nmol 92-105 C
Rh 0.8 pmol 0.3-1.0 Sc; N?
Pd 0.6 pmol 0.2-0.6 Sd
Ag 25 pmol 0.5-45 N; Cx
Cd 0.7 nmol 0.001-1.1 N
Sn 4 pmol 1-12 Dd?
Sb 1.2 nmol ? ?
Te 0.6 pmol 0.4-1.7 Sc
I 0.4 umol 0.2-0.5 N
Cs 2.2 nmol - C
Ba 100 nmol 32-150 N
La 30 pmol 8-57 Sd

Ce 20 pmol 16-26 Sd
Pr 5 pmol 1-8 Sd
Nd 25 pmol 5-40 Sd
Sm 4 pmol 1-6 Sd
Eu 1 pmol 0.3-1.7 Sd
Gd 6 pmol 2-9 Sd
Tb 1 pmol 0.2-1.5 Sd
Ho 2.5 pmol 0.5-3 Sd
Er 8 pmol 2-10 Sd
Tm 1 pmol 0.3-1.5 Sd
Yb 7 pmol 1.5-11 Sd
Lu 1 pmol 0.2-1.8 Sd
Hf 20 pmol ? Sd
W 56 pmol 45-67 C
Re 40 pmol - C
Pt 1 pmol 0.54-1.64 Sd
Au 50 fmol 20-200 variable
Hg 2 pmol 0.5-12 Cx;Sc
Tl 60 pmol - C
Pb 10 pmol 5-175 High in surface water
Bi 0.1 pmol <0.015-0.24 Dd
U 13.6 nmol +- 1.2% C

All values units per kg.
C = Conservative
N = Nutrient
Sd = Surface depletion
Dd = Depletion at depth
Mm = Mid-depth minima
Sc = Scavenged
Cx = Complex

From "An introduction to the chemistry of the sea", Michael E. Q. Pilson 1998
</pre>

[rshimek]

Hi Tatu,

<b><font color="deeppink">THANKS!!!</b></font>

I truly appreciate it. There is no problem posting any data as long as they are appropriately cited or referenced.

:D

[Randy Holmes-Farley]

Tatu:

That table has almost exactly the same data as the table that I have.

[rshimek]

Hi Randy,

Okay - I will work with these data in the downstream comparisons with trace elements.

:D

[simonh]

Thanks guys.

It was the last figure in your article which I found very useful for seeing which elements were higher / lower than NSW. It was just slightly confusing that the 'not detected' items were relabled as absent with small bars well below NSW. IMO it would have been nice to draw these bars from 0 upto the detection limit expressed relative to NSW. Maybe that was not done because these would be very high relative concentrations for some of the trace elements?

[rshimek]

Hi Simon,

I may try a graphic like you describe, but it is a bit dicey in construction - as detection limits vary over couple orders of magnitude; it maybe easier to do it with a table.

The main problem here is "pitching to the audience." While people like yourself, Randy and Tatu have no trouble working with a log-scaled graph, I think (based on the talks I have given) that a lot of folks do. If I could have figured out a way to present the data I did differently - and concisely - I would have.

And I should have use ND or some such abbreviation instead of "Absent" - but just didn't flash on it at the time.

:D

[Mutagen]

Ron,

A short while back the possibility of posting specific research papers pertinent to the hobby was raised. At that time, you suggested there may be copyright infringement problems and/or costs associated with that idea. However, you indicate posting data is OK as long as it is appropriately cited or referenced. I'm not saying that both statements can't be true, but I am curious as to where one draws the line. What are your thoughts on this?

Thanks for your time, Tracy Gray

[rshimek]

Hi Tracy,

The difference is one of posting whole articles for dissemination, versus a small portion, such as a table with appropriate citation.

In the first case, the whole article is copyright and widespread distribution is generally forbidden, although an individual is generally allowed to make a single copy for their own use. Generally, copying or distribution of single tables is allowed as long as there is adequate acknowledgement of the source.

In the latter case, you are using the information for comparative purposes. In the former, you effectively publishing the article without the permission of the copyright owner. Most books and journals will have a page which specifcally allows the latter and forbids the former.

The line is drawn when you start to publish wholesale excerpts from books or articles and then sell the material for a profit. This is generally considered a severe no-no....
:D

[leftovers]

Dr. Shimek,

Given the statements in this forum about the timeliness of the NSW values, would it not be prudent to run the assay comparisons against recent (also, the data, if not run in the same fashion or with the same technique could be significantly different) data to validate the findings? Or further, actually test NSW from several reef sources, as they are likely to vary significantly by locality? I would think given the scope of your study that you would have wanted several samples from GBR, Fiji, and Indonesian reefs tested by your peers or yourself for at least one geographically relevant NSW data point.

Another factor not addressed that could have had serious consequences for the analysis is filtration of the samples prior to analysis. I did read in another post that there was no filtration: wouldn’t a structured test protocol that eliminated this been more prudent? Or was this just a simple over site?

Given the supposition that data on reef water composition is lacking (or non-existent) and the known variance of reef keeping habits as you stated, doesn't that more than trivialize your conclusion that, "Well, even the most adventurous of these folks would likely not have guessed the differences found in these 23 reef tanks"?

I would think that this is not only a rather obvious conclusion but would further implicate that the absolute randomness of animals-inverts/verts’, live/dead rock, sand, the unknown food sources, and variability within water sources for makeup water as stated in your own words would prevent any kind of meaningful or statistically relevant conclusion to be drawn. I wonder as well, seeing as there is no mean data for reef NSW water values to compare these samples to, if any conclusion would hold value to the hobbyist or the scientist.

The dendrograms, while interesting pictures, show nothing relevant since the clusters (from what I can tell) don’t show any strong relational value. Further conclusions that the similarities are due to the wide spread use of Instant Ocean is also meaningless unless you know this to be fact, so any further inferences don’t hold any weight. I didn’t see any information about the samples themselves this would have been extremely helpful to know before hand in reading this report.

One final comment/question is why didn’t you use a 3D picture or cluster picture to display the 3D distribution data? I think doing so would have more clearly illustrated your point.

Thank you for your time,


Lefty

[rshimek]

Originally posted by leftovers

Hi,

Given the statements in this forum about the timeliness of the NSW values, would it not be prudent to run the assay comparisons against recent (also, the data, if not run in the same fashion or with the same technique could be significantly different) data to validate the findings?

Sure, and those comparisons will be in the next article.

Or further, actually test NSW from several reef sources, as they are likely to vary significantly by locality?

I'd love to do, you got the money to pay for it?

I would think given the scope of your study that you would have wanted several samples from GBR, Fiji, and Indonesian reefs tested by your peers or yourself for at least one geographically relevant NSW data point.

Right. Get real. You don't have any conception of the costs involved with such a testing program. If the data are not in the literature, and to the best of my knowledge they are not, just going to these sites to take the samples would cost several thousands of dollars.

Another factor not addressed that could have had serious consequences for the analysis is filtration of the samples prior to analysis. I did read in another post that there was no filtration: wouldn’t a structured test protocol that eliminated this been more prudent? Or was this just a simple over site?

It was no an oversight. I consider it of minor consequence. To do this work I would have needed have the water samples filtered at each person's house. The water would have need to be filtered to submicron levels to remove bacterial particulates. While this would have been feasible had I been doing all the samples in my lab, it is not feasible to expect volunteers to do this.

Likewise by the time samples got the labs, the state of any original bacterial or other particulate could have been significantly altered. If the samples had been filtered previously, more particulates could have been formed.

The lab doesn't do this procedure, it simply is neither required nor commonly done in marine evironmental analyses in that region. So they did not do it. I choose to assume that the relative amount of particulate material is fairly small, and will deal with it in the discussion.

Given the supposition that data on reef water composition is lacking (or non-existent) and the known variance of reef keeping habits as you stated, doesn't that more than trivialize your conclusion that, "Well, even the most adventurous of these folks would likely not have guessed the differences found in these 23 reef tanks"?

Frankly, I don't see your point.

I would think that this is not only a rather obvious conclusion but would further implicate that the absolute randomness of animals-inverts/verts’, live/dead rock, sand, the unknown food sources, and variability within water sources for makeup water as stated in your own words would prevent any kind of meaningful or statistically relevant conclusion to be drawn.

The tanks inhabitants are neither random nor unknown. Neither are food sources. Make up water is inconsequential, as I am discussing the final product.

I wonder as well, seeing as there is no mean data for reef NSW water values to compare these samples to, if any conclusion would hold value to the hobbyist or the scientist.

That is up to for the individual to decide.

The dendrograms, while interesting pictures, show nothing relevant since the clusters (from what I can tell) don’t show any strong relational value.

Actually the clusters show very strong relational values, I am sorry you can't discern that.

Further conclusions that the similarities are due to the wide spread use of Instant Ocean is also meaningless unless you know this to be fact,

That statement is incorrect.

I didn’t see any information about the samples themselves this would have been extremely helpful to know before hand in reading this report.

Perhaps. I doubt it. In any case presenting those data, about 140 pages of text, is impossible. You will have to be satisfied with the summary data.

One final comment/question is why didn’t you use a 3D picture or cluster picture to display the 3D distribution data? I think doing so would have more clearly illustrated your point.

Glad to do it. Send me the software to that can do it, and I will be glad to use it to produce an image. The three images I used clearly show the relationships of the samples to the confidence interevals and relevent points from the 3 primary axes of ordination. Data are presented this way in most journals, so we have to live with it here.

[npaden]

I think one very interesting point that Leftovers made that would have been a VERY good control standard would have been to have a couple people from around the world collect some water from a reef near them. I think a sample from the phillipines, GBR, Florida Keys would have been easy to obtain, and while I was unwilling to contribute $100 to get my water tested I would be willing to contribute $100 to get one of those samples tested to compare to your results. That would have also solved the below detection limits on your various tests as the control standards would have also been in the same category.

Just a thought, I'm not sure if you can do any more samples or not at this late date.

FWIW, Nathan

[rshimek]

Nathan,

The samples are actually not easy to obtain and recent work indicates that there may be significant along reef as well within reef differences.

However, the point is moot. This was a study of tank water, not reef water.

:D