What are common biases in sampling populations for genetics studies and how can they impact results?

• A. Non-random sampling, small samples, ascertainment bias, uneven sampling across populations; biases can distort estimates of allele frequencies, heterozygosity, Fst.
• B. Random sampling with large samples; biases improve estimates.
• C. Only sampling one locus is sufficient; biases do not affect results.
• D. Using only ancient DNA eliminates sampling bias.

Answer: (A)

Common biases in sampling populations for genetics studies include non-random sampling, small samples, ascertainment bias, and uneven sampling across populations. These issues matter because they distort the basic measurements researchers rely on to understand genetic variation.

Non-random sampling means you’re not getting a representative slice of the population. If you pick individuals from a limited locality or from related individuals, you’ll overrepresent certain alleles and genotypes, which can mislead conclusions about overall diversity and structure.

Small sample sizes amplify random fluctuations. Rare alleles can be missed, and estimates of allele frequencies and heterozygosity become unstable, cutting into the reliability of downstream inferences about population history or mating patterns.

Ascertainment bias happens when the genetic markers studied were discovered in a particular subset of populations or are biased toward common variants. This skews the site frequency spectrum and can lead to under- or overestimation of diversity, as well as distorted views of how different populations are related.

Uneven sampling across populations creates uneven statistical power and can make some groups seem more differentiated or similar than they truly are. This affects measures of population structure, such as Fst, and can misrepresent the degree of differentiation or gene flow.

Together, these biases can lead to biased estimates of allele frequencies, heterozygosity, and population differentiation, which in turn distort interpretations of population history and evolutionary processes.

Other options imply that biases can be neutral or corrected simply by using random large samples, sampling only one locus, or relying on ancient DNA to “fix” bias. In reality, random large samples reduce bias but don’t eliminate it; a single locus cannot capture genome-wide patterns; and ancient DNA introduces its own biases and limitations rather than automatically removing sampling bias.