Hi Garris

 I think it is very difficult to take representative samples of soil microbial communities for a large spiral scales for getting more accurate result. The soil microbial population vary in soil according to presence of organic matter, Soil pH, moisture, temperature etc. as well as presence of antagonistic or synergistic type of microbial population. Different agricultural practices like amount of different chemical of organic fertilizer application, method and amount of irrigation water applied, cultivation method........ etc and also different environmental parameters like humidity, sunshine, wind velocity........etc  play a significant role for presence of different microbial population in soil of a particular land. It is quite difficult to maintain same condition for all the parameters as mentioned above for a larger area. So, for getting more accurate result of most probable number of microbial population in soil ,it is better not to take representative samples of soil microbial communities for a large spiral scales.  

with regards.   

There is quite a lot of theory on the relationship between spatial variation and numbers of samples that is required to cover all that variation. Basically, it will be very difficult to know that you covered everything (what some people call " representative") that is there on a specific area. An approach could be to sample in a pattern to cover your area, and then use spatial statistics and look for the variogram of that data (Sill). For more details and good contact info: see the attached book.

Article Sampling for Natural Resource Monitoring Sampling for Natura...

And the direct link (there called chapter, but it is a full book)

Chapter Sampling for Natural Resource Monitoring

The best answer can be found in the experiment. If I were you, I would set up 2 experimental plots (1 x 1 m) in the same soil and climate conditions. The first one, I would divide into subplots 25 x 25 cm and take the soil samples from each of them. The second one I would use to take a mixed probe. After that - just to compare the results and to find the answer for your question... In addition, on the first plot you would be able to find some scale depended trends: the means for 25cm2, 50 cm2 and 1 m2.

Dear Dr Bedernichek, why choose 25cmx25cm? Why not 2x2cm? Or 25cm2? Microbes live on the (inner)surface of soil particles, so isn't 25cm an incredible far distance? I agree one just needs to test, but in the original setting, and taking as much pre-info as possible into account.

Dear Dr. Vorenhout,

That was just an example. No matter 10 x 10 mm or 10 x 10 m – such an approach gives the researcher an opportunity to find some scale-depended properties of soil, forest, community etc. You are absolutely right that 10 cm might be a long distance for microbes. However, as far as I understood the question of Dr. Heath W Garris, the problem was to describe microbial community of large territories. Therefore, a high level of generalization is needed. 

Sampling has been done to death on different scales for elements.  For example look at publications by Alex McBratney et al.

Hi Heath!

This problem has been solved in sedimentology by taken a number of random samples within the prescribed area (Krumbein 1934: Am J Sci 27). In principle, the larger the number of random samples, the smaller the probable error; e.g. 4 random samplers reduce the probable error by 50%, 10 samples by 70%, thereafter the gain becomes minimal. I see no reason why this should not also work in the case of other sediment components such as organic content and, yes, microbes, as long as the sediment (or soil) is reasonably homogenous. The paper by Krumbein (1934) is attached.

Burg

Collecting a composite samples is obviously one possibility.

Read: 134. Roger PA, Jimenez R, Santiago-Ardales S (1991) Methods for studying blue-green algae in ricefields: distributional ecology of the organisms, sampling strategies and estimation of abundance. Int. Rice Res. Paper Series n° 150, 19 pages. The International Rice Research Institute, P. O. Box 933, Manila, Philippines. 

Article Methods for studying blue-green algae in ricefields : distri...

Consider  alternative to simple random sampling designs - but designs that still provide soil (and therefore microbial)  samples with known probability (uniform probability designs - which include simple random and systematic - are often easier to understand. For example, the vegetation type, condition and coverage  of the terrain probably reflects many  environmental conditions , including soil fauna.  One could use some measure of such 'visible' information  as indicator variables for more targeted sampling designs that aim to obtain estimates of population totals or densities  of acceptable accuracy with fewer samples than a random 'sample'.