Friday, February 13, 2015

Surprising Mesh-Size Effects

Though I shouldn’t have been, I was surprised at the effects different sized sieves can have on the screening of fossil-strewn matrix.  Turns out, I seem to have some company among professional paleontologists, which is, itself, also surprising.

Here’s the source of my first surprise.  My previous post featured a picture of a plastic container with roughly one and a third cups of matrix.  For the past couple of weeks, I’ve been slowly working through that Miocene Epoch material in search of fossils (both macro- and micro-).  My early finds were mostly large fossils, shells and shark teeth primarily, spotted with the naked eye.

Then, as I worked deeper into the container, my finds shrank in size . . .

and, when the microscope swung into play, the finds were that much smaller, many shells from foraminifera and several from ostracodes.  Foraminifera appear in the image below (scale bar is 500 microns which is 0.5 millimeters).

What mesh size had I used originally to screen this material?

Well, it didn’t take long in this process to realize that this material hadn’t been washed or screened at all.  It was just as I’d gathered it, which means it was the equivalent of using a sieve mesh with no openings in it.  In the two systems of sieve mesh-size measurements with which I’m familiar, as the numbers assigned to sieves increase, the mesh sizes decrease (though not, I’m sure, to zero).  Here are a few examples from the U.S. Sieve Size and Tyler Mesh Size systems:

U.S. Sieve Size Tyler Mesh Size Opening in Millimeters Opening in Microns
No. 10 9 Mesh 2.000 2,000
No. 18 16 Mesh 1.000 1,000
No. 35 32 Mesh 0.500 500
No. 60 60 Mesh 0.250 250
No. 80 80 Mesh 0.177 177
No. 200 200 Mesh 0.074 74

(Tables offering more sieve sizes appear widely on the web.  One source is Appendix E:  Particle Size – U.S. Sieve Size and Tyler Screen Mesh Equivalents, in Fundamentals of Turbulent and Multiphase Combustion, by Kenneth K. Juo and Ragini Acharya, 2012.)

So, given no screening of this material, I shouldn’t have been surprised (but was) at what happened as I got down to the dust-like dregs of the matrix – the silica frustule of a diatom caught my eye, a kind of fossil I’d never found before.

The scale bar in the picture above was chosen deliberately because the smallest sieve that I own and use is a No. 80 in the U.S. Sieve Size system which has mesh openings that are 177 microns or 0.177 millimeters wide.

Yes, the frustules of these algae can be larger than what I found and, yes, they can be very abundant, but many are much, much smaller.  So, essentially, any fossils smaller than 177 microns, many of them presumably diatoms, that make it all the way through my stack of sieves (large mesh on the top, small mesh on the bottom) are never to be seen by me.  But, this issue is not mine alone.   Writing apropos of a non-paleontological context (quantitative analysis of extant marine ecosystems), marine biologist Kevin Zelnio noted:
The sieve size I use at the bottom, . . . , is the most important.  It is my cut off.  Essentially, I’m saying I’ll ignore anything that can fall through this size hole.  Ideally, this should be as low as possible, but I’m often limited by what sieve size my colleagues have used in past studies.  This is important because our results need to be compared to each other.  ((Sieve) Size Matters, EvoEcoLab, Scientific American blog, March 5, 2012.)
Seems so obvious – what can fit through the openings of the mesh in the final sieve eludes analysis.

Here’s my next surprise, another one that I should have anticipated because science is introspective and correcting.  Since, in paleontology, as well as in biology (. . . and in most of the rest of life), one may make decisions with significant consequences and sometimes not know or fail to keep those consequences in mind, there has developed a large body of (often strongly cautionary) literature analyzing the effects of mesh sizes on the findings of paleontological and biological research.

This is an important issue because sifting matrix through sieves is a core paleontological endeavor.  Indeed, one paleontologist has gone so far as to assert that “the real work of paleontology comes in extracting and identifying individual fossils from a bulk sample of rock or sediment (collecting bulk samples is as easy as digging a hole) . . . .”  (J. Bret Bennington, Confessions of a Statistical Paleontologist, Hofstra Horizons, Fall 2011, emphasis added.)  Though, certainly, not every paleontologist spends his or her time sieving samples, extracting and identifying the different fossil species contained in those samples, and counting individual specimens, many do, particularly those exploring the paleoenvironment of specific sites.  My efforts are not so ambitious.  Actually, I’m more of a tourist, visiting ancient environments to get a glimpse of what lived there.  I’m just passing through.

From my reading of just a fraction of the literature on mesh-size effects, the central (and not earth-shattering) conclusion is that what is kept and analyzed makes a fundamental difference for research findings.  In general (and not surprisingly), smaller sieve mesh sizes serve to increase, among other results, estimates of the biomass of extant organisms, the number of identified species, the number of individual specimens collected, and the retention of juvenile forms.  Yes, less is more.  All of this stands to reason given that species (fossil or otherwise) can vary in size, and juvenile specimens of a species are likely to be smaller than adults.  (A few papers from this literature are listed at the end of this post.)

Geologists Michał Kowalewski and Alan P. Hoffmeister have observed that “even studies that share virtually identical research goals and target the same groups of fossils collected from the same time intervals can vary greatly in mesh size.”  (Sieves and Fossils:  Effects of Mesh Size on Paleontological Patterns, PALAOIS, Vol. 18, No. 4/5, October 2003, p. 460.  This is hiding behind a paywall, sorry.)  To demonstrate mesh-size effects, they used a detailed database describing a large number of fossil marine benthic mollusks from two Miocene Epoch sites and simulated the use of different mesh sizes on the sorting of these fossils.  Certain paleontological patterns that have been studied over the years were affected by changes in mesh sizes.  For example, they found that, as mesh sizes increased, bivalve representation in the selected population decreased, and the percentage of mollusks with predator holes increased as did the percentage of encrusted specimens.  They concluded, “What matters here is the observation that if the Miocene dataset were originally obtained with coarser sieves, different results would have been reported . . . .”  (p. 465)  Though some species-diversity measures seemed to have been little affected by mesh sizes for this particular dataset, there were differences in the impact of changes in mesh sizes between the two sites that provided specimens to the dataset.  “The new insight offered by this analysis is that ensuring that a standard mesh size is used in comparative analyses is insufficient – the outcome may depend on the choice of that standard.”  (p. 465)

There are trade-offs, of course.  Keep too much and the analytical tasks may exceed the resources researchers have available to them (I know, I still a lot of the fine lees from that container of matrix to go through).  It’s also true that, in some instances, finer mesh sizes may negatively affect the analysis (see, for instance, Susan Kidwell’s paper listed below).

Mesh-size effects – often surprising, and, usually, they shouldn’t be.

Selected Additional Literature

John D. Gage, et al., Sieve Size Influence in Estimating Biomass, Abundance and Diversity in Samples of Deep-Sea Macrobenthos, Marine Ecology Progress Series, Volume 225, 2002.

Susan M. Kidwell, Mesh-Size Effects on the Ecological Fidelity of Death Assemblages:  A Meta-Analysis of Molluscan Live-Dead Studies, Geobios, Volume 35, 2002.

Antoine Morine, et al.,  Sieve Retention Probabilities of Stream Benthic Invertebrates, Journal of the North American Benthological Society, Volume 23, no. 2, June 2004.  (Hidden behind a paywall.)

Frank Peeters, et al., A Size Analysis of Planktic Foraminifera From the Arabian Sea, Marine Micropaleontology, Volume 36, 1999.  (Another paywall, it would appear.)

No comments:

Post a Comment

Nature Blog Network