The fall armyworm, Spodoptera frugiperda (J.E. Smith), is comprised of two genetically distinct strains that are morphologically identical yet exhibit differences in their behavior and physiology (C-strain and R-strain). Evidence of ongoing genetic differentiation between strains highlights the importance of considering strain identity in research and management of fall armyworm populations, but the logistical and technical burden of genotyping limits strain-specific applications. Controlled experiments with laboratory colonies have shown that the strains engage in allochronic (“allo” – different, “chronic” – time) mating behavior, with C-strain mating early in the evening (0–5 hours after sunset) and R-strain mating late in the evening (5–10 hours after sunset). Using temporal field collections and genotype data, we show that strain-specific variation in allochronic male mating behavior occurs across Texas and Florida fall armyworm populations, both of which act as primary source populations for annual migrations of this pest into the continental United States. Time of capture in pheromone traps was significantly different between strains in both Texas and Florida, with the R-strain males consistently being collected in the traps late in the night. The C-strain males were generally captured earlier in the night than their R-strain counterparts, though there was notable variation in the timing between nights and across locations. Allochronic behavior in field populations is consistent with previous laboratory studies reporting differences in the timing of mating between the strains, however increased variability in behavior within and across native populations was observed. Although allochronic behavior in local populations may partially contribute to reproductive isolation between the strains, the behavior is not consistent enough to serve as a complete reproductive barrier. Furthermore, the observed variability in behavior both within and between independent sampling events, especially in the C-strain, poses a challenge to the development of models that utilize time of capture as a predictive phenotype for monitoring strain identity in local populations.