I am thinking about doing SEM-EDS images in plant samples to verify metal distribution and accumulation. In both roots and leaves (mostly checking metal accumulation in trichomes). But I am unsure of how many biological and technical replicas would suffice to infer anything about the plant patterns of metal distribution. The papers I've been reading really don't specify that, even after metal quantification, or even after statistical analyses. How would you do it? Is it the standard minimum of 3 biological replicates? Or 1 biological replicate with several images? Appreciate any opinion on this.
I assume under "replica" you mean "specimen".
For publishing in bottom-feeding journal you need almost nothing, and one specimens will be more than enough.
For more presentable journal you need more specimens. At least 3 controls and 3 of interest. If you are going for statistics, then after measuring 10 specimens ask a statistician how many specimens (in addition to 10) do you need.
Vinicius Henrique De Oliveira It depends on whether you want your results to be representative of the bulk or not. Do you want to take a specimen or a sample?
3 has no statistical validity whatsoever - throw away the 2 outliers and you're left with a perfect result. The term to look up on this site and on Google plus standard text books is 'standard error' which is inversely proportional to the square root of the number of experiments, determinations, particles etc. If, for example, you want 1% standard error, you'll need 10000 determinations...
Thank you both for the replies. Just to clarify a bit what I meant. By "replicas" I meant replicates (sorry, it got auto-corrected). So, my experiment has 12 treatments, with 8 replicates each, so in total I have 96 plants (96 pots). I would like to see the metal accumulation pattern in the leaves, and see if it varies across treatments... so how many images do I need to be able to infer anything? For instance, papers usually will give you 2 or 3 images of a leaf and then discuss it. But how many pictures/leaf samples/replicates they assessed in order to choose the representative ones published in the paper? It's rarely mentioned in the text.
Vinicius Henrique De Oliveira Again you have to decide what your required or specified standard error is. If you're not familiar with this term then grab a basic statistics book or look on the Internet. If it's 5% SE, then you'll need 400 images. If your ~ 100 plants are truly random then you'd need 4 random images/plant and analyze all 96.
However, I don't know what you mean by 'I would like to see the metal accumulation pattern in the leaves'. How do you assess this? How do you know it's different from leaf to leaf?
Three can have a huge "statistical" value. If 3 controls show 0 and 3 treated show 100, you are OK. Just do not mention statistics in your paper. But anyway it's not your case.
It is impossible to answer your question now, you need to acquire some data to see how big (if any) is difference between treatments, then you may have some idea about needed quantity of specimens. It looks like you need help of a statistitian, not SEM-EDS specialist. Just remember you do not need to achieve some level of some standard error, you are interested in statistical significance in change of state of your specimens. Hope you do not have too many independent variables.
thank you both for the replies. But what I was really looking for is an opinion on the technique, and what is considered a decent result. For example, if I harvest, dry and weigh 3 plants of each treatment, I know I'll get a biomass result. But what happens when I am doing SEM imaging? What is usually accepted as result? I cannot take SEM images of all leaves of my 96 plants to get an idea of how much it varies, it is probably a lot and I don't have time and resources for that. On the other hand, one snapshot of 1 single leaf seems too little in order to be able to say anything. So that's why I was hoping to get some guidance on this. I see a lot of papers on this and I never see like "we took 3,000 images, and this is what we got".
Hi Alan F Rawle , the SEM-EDS technique can determine metal concentrations in selected areas or points in a given leaf. For example, guard cells. So you can quantify metals in specific structures. It does probably vary a lot from leaf to leaf, from cell to cell etc. That's why I wanted to know how people are confident enough that the image published in the paper is of any significance (how did they choose images, and out of how many)
Vinicius Henrique De Oliveira Here's a reference from NBS (that became NIST in 1987):
The National Bureau of Standards has stated that at least 10,000 separate images (not particles) need to be measured for statistical accuracy in image analysis
'S(tandard error) is proportional to N-1/2 where N is the total number of particles measured…This consideration implies that image analysis may require the analysis of on the order of 10000 images to obtain a satisfactory limit of uncertainty' (p 718, paragraph 1) A L Dragoo, C R Robbins, S M Hsu et al “A critical assessment of requirements for ceramic powder characterisation” Advances in Ceramics Vol 21 Ceramic Powder Science (1987). The American Ceramic Society Inc. (pp 711 - 720)
Vinicius Henrique De Oliveira I still need to know how you differentiate differences in metal concentration... What is a valid difference for you?
Hi Alan F Rawle , so that's the thing. When we're doing metal concentrations in a macro scale, let's say "plant shoots", I would digest and determine the metal concentrations of at least 3 plants (replicates) for each treatment (then run an ANOVA or something). But when we're talking about microscopic images, it's hard for me to grasp, how many images/leaves/plants would be enough. Because in one single image you can take several x-ray measurements. Ideally it would be good to assess every single leaf of every single plant, but since that's not possible, I wondered how many would suffice for a valid SEM result.
Vinicius Henrique De Oliveira I’m not sure that we’re on the same page here. Enough images for what purpose? What do you mean by a ‘valid SEM result?’ A picture is a picture. What part of the whole is imaged? The entire leaf?
Alan F Rawle Let me give you this example. The paper attached has an SEM image (Figure 3), in which metal contents have been determined in particular points of the leaf. For instance there is a high Mn content in the base of the trichome (n. 7). And that was it. All I want to know is how many SEM pictures one does usually take in order to select 1 for the paper. I'm pretty sure it wasn't 10,000, but I think it wasn't the first one they took either. That's why I was asking people that work with SEM of plants, to see what they usually do.
Vinicius Henrique De Oliveira I'll answer separately by PM with a word document.
- Badges
- Science method