More than 50 years ago, Ledyard Stebbins (1971) proposed that polyploid species are generalists that can tolerate a wide range of environmental conditions. However, little has been known about the underlying mechanisms of environmental robustness of polyploid species. Using a model allopolyploid species Arabidopsis kamchatica, we developed bioinformatic tools for genomic analysis, and PlantServaton, a machine learning-based phenotyping method in natura, or in naturally fluctuating environments. To test whether the allotetraploid A. kamchatica inherited and combined the environmental responses of the two diploid progenitor species A. halleri and A. lyrata, we grew these species and A. thaliana for three seasons at two field locations. We analyzed > 4 million images (12 genotypes × 20 replicates × 2 sites × 16–24 images/day × 150 days/year × 3 years). We estimated anthocyanin content as a proxy of stress, and examined environmental and genotypic effects on it. Anthocyanin content was affected by past radiation, coldness and precipitation in the naturally fluctuating conditions, consistent with the previous studies in regulated chamber conditions. Synthetic polyploids combined responses of diploid progenitor species and recapitulated the response of natural polyploid genotypes. These data support the combined environmental responses of polyploid species.