The concept of electromagnetic signaling in biological tissues has been around almost since Hertz's experiment proving the existence of electromagnetic waves. Although electromagnetic signaling would better explain the speed and precision of the work of the genome and coordination between the cells than chemical signaling, molecular mechanisms and experimental evidence for it are insufficient. Recently, we proposed that DNA, being the central program of life and a very stable substance, is directly involved in electromagnetic signaling and that its sequences harbor sequence-dependent electromagnetic oscillators. We also proposed two types for sequence-dependent electromagnetic oscillation - the ones that include delocalized negative and positive charges. The first type is proposed to occur in purine stretches in DNA, i.e. stacks of purine bases (A and G) with electron clouds delocalized within stacked aromatic rings. Here, we utilized the public data on genetic variations in genomes of several biological species to test the evolutionary pressure on the length of purine stretches. The single nucleotide polymorphism (SNP) replacing a purine with pyrimidine would disrupt the purine stack and the corresponding delocalized electron cloud and thus impact its oscillation frequency. As a result, it was demonstrated that there is a consistent evolutionary pressure towards the elongation of purine stretches in the genomes of the human and all other tested species. Additional analysis demonstrated that this pressure is independent of the bias caused by unequal chemical nucleotide lability and is therefore genuinely caused by the functional evolutionary advantage of longer purine stretches. This offers additional support for the proposed mechanisms of electromagnetic signaling in the genome.