Why we sample.
Recording every individual in a population is impractical, unnecessary, and expensive (Magurran 1988). Instead community ecologists and scientists in general take replicated samples to represent the overall community.

How many samples?
Too few samples can yield inaccurate and misleading results; too many becomes cost prohibitive (Eckblad 1991). Sample number is influenced by available resources, ethical concerns, and required accuracy. Detecting differences in parameters among sample groups depends upon parameter variability (Figure 1), and the size of the difference one is trying to detect (Gotelli & Ellison 2004). Gotelli & Ellison (2004) suggest the Rule of 10 as a minimum sample number per comparison category or treatment. Importantly, they point out that this rule cannot be universally applied, particularly with large-scale experiments.