Dr. Rob Dillon, Coordinator





Thursday, September 3, 2015

The Lost Thesis of Samantha Flowers

Warning.  The essay below is the fifth in a five-part series on species relationships in the enigmatic North American “stagnicoline” lymnaeids.  I will assume that you have read all four of my previous posts.  In addition, I also make explicit references to two essays in my 2012 series on this same subject, [10May12] and [4June12].  In fact, it would probably help if you started with my [20Nov06] essay on F. C. Baker, and read my [28Dec06] essay on the classification of the Lymnaeidae as well.  Stand back, I’m going to try Science! [1]

Subsequently published as Dillon, R.T., Jr. (2019b)  The lost thesis of Samantha Flowers.  Pp 95-106 in Freshwater Gastropods of North America Volume 2, Essays on the Pulmonates.  FWGNA Press, Charleston.


Insofar as I was able to determine from my vantage point 820.64 miles south of Ann Arbor, Samantha made good progress on her thesis research through the 2012-13 academic year, and into the field season following.  I was pleased to see the abstract of a talk she gave at the AMS meeting in The Azores in July of 2013, although I myself was unable to attend.  And on 15Aug13 I received a very upbeat email from her, my first in over a year.  She reported that she was “currently on the last leg of my Master's journey, prepping a manuscript to recount my arduous tale of stagnicoline systematics that should be wrapped up within the next month or two.”  She also promised to keep me posted “for when the sequences are thrown up on GenBank [2].”  And that was to be the last I ever heard from her.

As that “month or two” stretched into 2014, with no reply to my repeated emails, I began to worry that something might be amiss with our promising young malacologist.  Googling around on the University of Michigan website, I was able to confirm that Samantha did indeed defend her thesis, “Inferences into species delimitation of Nearctic Stagnicola using geometric morphometric and phylogenetic methods,” on November 15, 2013.  The outcome I was unable to determine.  But surely, I thought, if her thesis were successfully defended and signed, it must ultimately appear for download (or purchase?) through some public outlet somewhere, yes?

No.  After more than a year of watchful waiting, in January of 2015 I finally emailed her major advisor, Tom Duda, to inquire about the fate of Samantha and her thesis.  Tom confirmed that Samantha’s 2013 defense was indeed successful, and that he himself was surprised not to find her thesis uploaded to the University of Michigan’s “Deep Blue” server.  Apparently The University does not have firm rules regarding the deposition of MS theses.  And Tom further confessed, “In the past we have requested that theses and dissertations be deposited in our Mollusk Division library, but regret that it was my oversight (in combination with her rapid departure and her not responding to emails after she left) that got in the way of this happening with Samantha.”

And in fact, as our conversation developed, it materialized that Tom did not have a clean, final copy of Samantha’s thesis himself.  He had apparently returned his only copy to her with written comments, and she disappeared.  I suggested that he might check with some of the other members of her committee, and he was able to locate a “near final form” version which he shared with me in April.  But Tom has asked me not to distribute the document any further, since the version from which I am working still has some errors.

Samantha’s Thesis [3] is a blockbuster. Her results simultaneously reinforce a large and growing body of research confirming the dramatic ecophenotypic effects of habitat on freshwater gastropod shell morphology, and shatter 200 years of set notions about systematic relationships in the North American stagnicolines.  Let’s digest her work in five steps.

First, Samantha’s CO1 sequence data suggest two biological species.  Perhaps some of you will recall the review of interpopulation sequence divergence in L. stagnalis I posted on the FWGNA blog in [4June12].  There I argued that the general rule-of-thumb estimate of 5% CO1 divergence often observed among biological species of pulmonate snails seems applicable to within-continent comparisons of lymnaeid populations worldwide.  This is not a law, it is a very broad-brush guideline [4].

So below I’ve reproduced a (rather heavily-edited) version of Samantha’s “Collapsed Bayesian-inferred CO1 tree,” with state and province abbreviations marking samples from ME = Maine, MI = Michigan, MB = Manitoba, and so forth.  Setting aside the single Lymnaea arctica sequence that Samantha mined from the Barcode of Life Database [5], the ABGD prior max distance bars at right seem to suggest the two clusters of stagnicolines I have labelled “V1” and “V2”.  Although the pairwise sequence differences between these two groups apparently do not consistently reach 5%, an eyeball estimate from Samantha’s scale bar, together with the plot of pairwise genetic distances Samantha offered elsewhere in her thesis, suggests to me that they probably often do.

Note that I have modified the noun “species” with the adjective “biological” here.  This is because populations of the two putative species seem to occur sympatically, at least in some cases.  More under my fourth point, below.

Second, the distinction between these two putative biological species does not coincide with taxonomic tradition, as historically based on shell morphology.  Samantha classified each of the individual snails she sampled for her C01 analysis using geometric morphometrics, digitizing their shell outlines with the large set of sliding landmarks [6] shown in the colorful figure I have reproduced at the top of this essay [7].  She recognized four nomina by shell fatness – identifying the green and red categories as emarginata, the gold as elodes and the blue as exilis.  The gold was an unfortunate color choice – nearly invisible between the red and blue in her figure. 


But in any case, the correct way to define any of these nominal taxa would have been by reference to populations sampled from their type localities.  God Knows I Tried to help Samantha with this critical component of her thesis, but for a variety of reasons, it just didn’t work out.  So I have deleted Samantha’s specific names from the CO1 tree above [8] and substituted simple color coding according to her morphometric analysis [10].  And it will be obvious that the four color categories do not correspond to the two putative biological species.  Cluster V2 shows all four colors, and cluster V1 shows three of the four.

Third, the putative biological species do coincide with Brady & Turner’s V1/V2 taxonomy.  Here I’ll ask you to open my essay of [10May12] in a separate window, and refresh your memory regarding Brady & Turner’s “cryptic stagnicoline” populations from NW Pennsylvania.  Although all four of the B&T populations inhabited fishless marshes, and all four bore dark, skinny shells typical of elodes, their “Hartstown Marsh” population demonstrated a consistently larger (and perhaps more “flat-sided”) body whorl than their Conley, Osgood, and Killbuck populations.  Kip Brady’s common garden experiments suggested that this body whorl difference seemed to be heritable.

Brady & Turner [11] considered that their Conley, Osgood, and Killbuck populations demonstrated “typical” L. elodes shell morphology, and called them V1.  They called their Hartstown Marsh population V2.  See note [12] for an interesting story about the example specimens figured at right below.

I forwarded samples of all four B&T populations to Samantha in July of 2012 [13].  And sure enough, samples from the Hartstown population (marked as PA-h on my version of Samantha’s CO1 tree) appeared genetically distinct from the Conley, Osgood, and Killbuck samples (marked PA-c,o and PA-k).  So although subtle, there does appear to be a shell morphological correlate to C01 sequence divergence between the two putative species.  The key character does not seem to be the traditional fat/skinny dichotomy, but rather the relative size of the body whorl [14].

Fourth, evidence suggests that the traditional taxonomy of North American stagnicolines may have been based on shell characters largely ecophenotypic in their origin.  The best example, ironically, comes from Douglas Lake, the home of the University of Michigan Biological Field Station.  Samantha sampled 4 individuals from the waters of Douglas Lake itself, all of these being classified as “emarginata-ovate” by her morphometric criteria, which I have marked with green letters “d” in the C01 tree above.  This small sample included three individuals belonging to putative biological species V2, and a single individual belonging to putative species V1.  Samantha also sampled 8 individuals from “Douglas Lake Pools,” presumably marginal ponds not directly connected to the lake itself.  All of these individuals were classified as exilis by Samantha’s morphometrics, and are marked with blue letters “d” above.  This included 6 individuals classified as putative species V2, and 2 classified as species V1.

Thus Samantha’s data suggest that two biological species of stagnicoline lymnaeids seem to co-occur sympatrically in Douglas Lake, both bearing fat shells of emarginata morphology in the main lake, and both bearing skinny shells of exilis morphology in marginal pools.  We search the world over, and sometimes the answers we seek are right on our own doorsteps.

And fifth, we do not actually know the correct names for either of the putative biological species.  Here I must pause, and wipe a tear from my eye.  For some reason known but to God, Samantha did not sequence that sample of topotypic L. catascopium I gave her in June of 2012.  Was this tragic oversight related to some sort of funky decision-making late in her research, regarding the taxonomy to be employed in her thesis as a whole?  See note [8] below for more.

In any case, as I have repeatedly emphasized (to Samantha, and to you all as well!), catascopium (Say 1817) is the oldest name available for any North American stagnicoline population.  One of Samantha’s two putative biological species almost certainly must be catascopium by definition, and the correct name of the other species depends.

So what to do?  Almost all of Samantha’s pale/fat snails, which might conventionally be identified as catascopium, were classified as V2.  This set included the sample I sent her from Maine, graphed as a big green triangle at the top of her C01 tree.  And almost all of Samantha’s V1 individuals demonstrated the dark/skinny shell morphology conventionally associated with elodesSo let us provisionally call the V2 species Lymnaea catascopium, leaving the name Lymnaea elodes for putative species V1.  This is admittedly a judgement call, but seems most consistent with the taxonomy currently employed by workers in the field.

Have I beat this horse long enough?  Let me conclude with two recommendations for further study.  First, the hypotheses advanced here can be tested with a good genetic survey of the stagnicoline populations inhabiting the Douglas Lake area.  Somebody needs to use microsatellites, or old-fashioned allozymes, or even older-fashioned breeding studies, to test the hypothesis that two reproductively isolated stagnicoline species are sympatric in that lake, not corresponding to the traditional fat catascopium / skinny elodes dichotomy, but rather corresponding to the new V1 elodes / V2 catascopium dichotomy.  And second, somebody needs to go back up the Delaware River and fetch us some more topotypic catascopium.  And find us some topotypic elodes at Lake Canandaigua, while shopping around in Yankeeland for lymnaeids anyway.  Not it.


Notes

[1] This catch phrase comes from the online comic, xkcd.com.  And although the xkcd logo shows a stick-figure scientist flamboyantly flourishing a beaker and a calculator, real science is at least as much theoretical as applied.  To be quite precise, science is the construction of testable models about the natural world.  This essay is real science.  Stand back.

[2] Not only did Samantha ultimately fail to make her MS thesis available from any public outlet, she also failed to “throw up” any of her sequence data on GenBank.  Alas.

[3] Flowers, S. L. (2013)  Inferences into species delimitation of Nearctic Stagnicola (Gastropoda: Lymnaeidae) using geometric morphometric and phylogenetic methods.  M.Sc. Thesis, University of Michigan, Ann Arbor.

[4] Two disclaimers.  First, gene trees are NOT species trees!  They are weak, null models of population relationship.  And second, there is no cut point for species-level sequence divergence that isn’t more exception than rule.  See, for example:
  • Phylogenetic Sporting and the genus Laevapex [20July07]
  • Gene Trees and Species Trees [15July08]
  • The Snails The Dinosaurs Saw [16Mar09]
  • What is a Species Tree? [12July11]
[5] Lymnaea arctica was one of the (only approximately ten) North American lymnaeids that Hubendick (1951) considered specifically distinct.  I wish I knew more about the taxon.  Might arctica demonstrate genetic affinities with the Old World palustris/corvus/fuscus group?  But I went over to the “Barcode of Life” database to see if I could find the PPCHU85 arctica sequence that Samantha analyzed in her CO1 tree, and came up dry.  Very frustrating.

[6] Samantha really should have digitized more than the shell outlines.  In particular, the relative sizes of the shell whorls, especially the body whorl, seem to contain a great deal of heritable information in freshwater gastropod populations [14], which may be difficult to recover without landmarks on the suture lines or aperture.  See for example:
  • Dunithan A, Jacquemin SJ, Pyron M (2012) Morphology of Elimia livescens (Mollusca: Pleuroceridae) in Indiana, U.S.A. covaries with environmental variation. Am Malac Bull 30:127–133.
  • Dillon, R. T., S. J. Jacquemin & M. Pyron (2013)  Cryptic phenotypic plasticity in populations of the freshwater prosobranch snail, Pleurocera canaliculata.  Hydrobiologia 709: 117-127.  [PDF]
  • Dillon, R. T. & S. J. Jacquemin (2015)  The heritability of shell morphometrics in the freshwater pulmonate gastropod Physa.  PLoS ONE 10(4) e0121962. [PDF]
[7] Samantha’s original Figure 5 had a nasty little error in its legend, which I have fixed by deleting her legend in its entirety.

[8] Although Samantha did not explicitly cite any reference works to support her taxonomy, it is my impression that her choices of the category names emarginata-ovate, emarginata-narrow (“canadensis”), elodes and exilis follow the 1992 work of Burch & Jung [9]. Whatever the origin, her taxonomy is most unfortunate.  For unexplained (indeed unexplainable) reasons, Samantha seems to have dropped the oldest specific nomen available for stagnicoline lymnaeids, catascopium (Say 1817), from her methods and results sections in favor of emarginata (Say 1821).  But (I’m guessing here) the decision may have come late in her project?  Because a couple samples remain identified as “catascopium” in her draft Table 1, with purple coding in her draft Figures 10, 11, and 12.  What a mess.

[9] Burch, J. B. & Jung, Y. (1992) Freshwater Snails of the University of Michigan Biological Station Area.  Walkerana 6(15): 1 – 218.

[10] It is my broad-brush impression, based on nothing more than inspection of the figures in Samantha’s draft thesis, plus the sets of stagnicoline shells figured by Burch and Jung, that samples our Michigan colleagues tend to call emarginata-narrow (“canadensis”) and elodes may tend to represent putative species V1, and the samples our colleagues call emarginata-ovate or exilis may tend to represent V2,  But since Samantha didn’t digitize the shell aperture or suture lines, the distinction was not recovered by her morphometrics.  So let’s just focus on Samantha’s color coding, and set her taxonomy aside to the extent possible.

[11] Brady, J. K & A. M. Turner (2010) Species-specific effects of gastropods on leaf litter processing in pond mesocosms.  Hydrobiologia 651: 93-100.

[12] For several weeks during the spring and early summer of 2012, I held cultures of the four B&T populations here in Charleston, dissecting samples to hunt for anatomical distinctions that I ultimately did not find. During that period both the Killbuck (V1) and Hartstown (V2) populations laid eggs.  These I hatched and reared for quite a few months in my standard plastic aquarium boxes, at densities that were certainly too high, largely neglecting them, changing their water infrequently.

In any case, my results seem to confirm those of Kip Brady.  My standard culture water here in Charleston is almost certainly much softer than that to which stagnicoline lymnaeids are usually adapted, yielding the chalky appearances of the two example shells figure above.  Yet the V1 offspring did indeed seem to develop relatively smaller body whorls than the V2 offspring.

[13] And here’s another little confusion.  Unknown to me, Kip Brady sent Jack Burch samples from a couple of his stagnicoline populations several years prior to my shipment to Samantha.  Kip never heard anything further.  But apparently 5 of Kip’s earlier samples were sequenced at some point along with 8 of the individuals I sent to Samantha in 2012, and all 13 appear graphed side-by-side in Samantha’s C01 tree, under two different labelling schemes.

[14] Body whorl differences of this subtle sort seem to be quite heritable in freshwater pulmonates as a general rule.  In fact, body whorl differences were the way we initially distinguished Physa carolinae from Physa acuta back in 2009 [15].  For more, see:
  • The Lymnaeidae 2012: A clue  [9July12]
  • The heritability of shell morphology in Physa h^2 = 0.819  [15Apr15]
[15] Wethington, A.R., J. Wise, and R. T. Dillon (2009) Genetic and morphological characterization of the Physidae of South Carolina (Pulmonata: Basommatophora), with description of a new species.  The Nautilus 123: 282-292.  [PDF]

No comments:

Post a Comment