population variation in gene expression

[coutellec]

hello everyone,
I plan to use RNAseq to compare differential expression (between benign and stressful conditions) among populations that are genetically differentiated. Is it correct to pool biological samples per population (individuals representative of different families within populations)? If i do not pool families, then the response will not be representative of the population. On the other hand, mixing different genotypes may lead to interfering effects within a library (if mixed genotypes have opposed responses).
Thanks in advance for any advice,
Marie-Agnès