In this article we explore the concept and implications of three-dimensional (spatial, temporal, and genetic) in-field crop diversification to inform systems redesign towards ecological intensification. We first present a conceptual framework for classifying diversity in arable contexts. We then apply the framework to analyse two long-term systems experiments in The Netherlands where spatial and genetic diversity measures were implemented via strip and mixed intercropping with the aim to increase ecosystem service delivery: incidence and spreading rate of late blight (Phytophthora infestans) in potato (Solanum tubersosum L.), and biocontrol control potential in wheat (Triticum aestivum L.). In the case of late blight, potatoes planted in strips had significantly lower disease incidence than the monoculture reference across all years, and adding cultivar mixing within the strip was more powerful in mitigating late blight than spatial diversification alone. In the case of biocontrol in wheat, strips supported significantly larger (for all but one taxonomic group) and significantly more diverse epigeic natural enemy populations than the sole culture reference in all years. However, the addition of species mixing within strips did not further increase biocontrol indices compared to sole-wheat strips. These results imply that compromises between management complexity and ecosystem service enhancement are achievable through strip cropping, an operable practice with current machinery, and one that does not require a thorough reconfiguration of the production system. The three-dimensional diversity framework proved useful for unpacking experimental outcomes in terms of diversity-mediated mechanisms, however it requires further development before it can be used to facilitate multi-objective optimization.