Wednesday, 9 April 2014

Tree data in

I had a hypothesis a tuthree weeks ago about how trees of different species might abstract particular minerals from the soil perhaps as a consequence of the fact that one was deep-rooted and unlikely to fall over (Fraxinus excelsior, ash) , while another was merely skimming the surface of the soil and would heave itself prostrate in a light breeze (Crataegus monogyna, whitethorn).  My environmental chemists, fresh from three months of independent research using all the kit and instruments in the Environmental Lab, we well able to to measure the concentration of some key elements in plants tissue from the two different species of tree.  To keep things as equal as possible, I suggested that we analyse shavings of bark off living twigs. There are no leaves yet. If you take leaves from one species and seeds from another you don't know which variable is driving any changes you might observe.  An alternative hypothesis might be that plants sample their environment more or less at random sucking up whatever happens to be in the soil or dissolves in the water near the roots. And we had a "null hypothesis" which was that there may be differences in the concentration but they are just tracking random noise in the data and there are no consistent differences among the samples.

It is also standard scientific practice to get several replicates of the data.  If these are internally consistent you may be confident that you have measured something accurately and reproducibly.  You also need to boil the material up in concentrated acid with hydrogen peroxide to get the minerals you intend to measure into a solution that can be sucked into a delicate scientific instrument.  You can't just lay a billet of wood on the lab bench and expect the machines to deliver results.  We used two instruments - a Flame Emission Spectrophotometer (FES) for Lithium, Li; Sodium Na and Potassium K and an Atomic Absorbance Spectrophotometer (AAS)  for Iron Fe: Copper Cu and Zinc Zn.

I would have preferred to get replicates in triplicate but the microwave digester can only take 10 samples at once. We finished up comparing a number of different variables but with only two replications (less than ideal) of each condition: (two species) x (two samples each) x (two locations) x (two digestion methods) = 16 samples. We can conclude from this very preliminary study that Li and Na are essentially the same in all samples although Na is about 10x more common in plant tissue than Li.  The same for copper, although one sample seemed to be anomalously high as if someone had buried a cache of pennies under that tree.  Hmmmm, must get metal detector. Iron, Zinc and Potassium, contrariwise, seem to show a consistent pattern with high values alternating with low:
Sample Fe Zn K
Fr.ex1 32.8 33.4 27.9
Fr.ex2 52.2 164.2 8.8
Cr.mo1 24.0 35.8 25.8
Cr.mo2 72.6 151.5 7.5
But WTF!?  The results say that one of the Fraxinus samples is consistently like one of the Crataegus, and the other Fraxinus sample (cut from the same root-stock) looks more like the other Crataegus than it does to its genetically identical twig-mate. Dang! That writes the whole experiment off because the most likely explanation for the anomaly is that someone has made a cock-up on labelling the test-tubes. None of the students seemed to be particular upset by this turn out but if that sort of thing happened twice in the commercial or administrative (EPA, say) world, you would get your cards the next week. No matter what your technical competence diluting samples accurately or getting reliable results from a complex piece of kit; if you're not sure what you've measured it's ALL A WASTE OF TIME. The audit trail must be seamless.

Not totally - the results are sufficiently intriguing that we can take this forward next year perhaps with a final year project student.  And we need not measure Copper, Lithium or Sodium in future, there is no signal there.

No comments:

Post a Comment