Sampling error size vs population size

Question

Reading a methodology for reporting plant/seedling survival I found the following statement:

In small plantations or very small strata (1 to 3 hectares) the sampling error size is usually high and, therefore, a higher number of sampling sites is needed to obtain representative estimations (Spittler, 1995)

My question is, why is sampling error size high in smaller plantations/populations? I'm hoping for an answer with a real-world example (not mathematical definitions); I've found this to be in agreement with sample size estimates, where sample size is not linearly related to population (evident from the formulas). The rather unavailable/outdated reference is below.

SPITTLER, P. 1995. Guía Técnica para el Inventario Rápido de Bosques Secundarios en la Zona Norte de Costa Rica. COSEFORMA. Alajuela, Costa Rica.

Answer 1

This sounds to me like a version of the Central Limit Theorem. Essentially, if the plot you look at is small, there might only be a few plants, so the survival rate is going to be susceptible to large swings due to chance. At one extreme, say there were only 10 seedlings. A difference of two additional ones surviving just by chance could shift you from 4/10=40% survival rate to a 6/10=60% survival rate.

If you have a large plot with significantly more plants (say 1,000), you can be more confident in your results - i.e. a smaller error from the true value due to the fact that you're looking at a sample. There will still be variation between plots due to chance, but you're unlikely to result in variations like the 40% to 60% swing above.