We present a crop simulation environment with an OpenAI Gym interface, and apply modern deep reinforcement learning (DRL) algorithms to optimize yield. We empirically show that DRL algorithms may be useful in discovering new policies and approaches to help optimize crop yield, while simultaneously minimizing constraining factors such as water and fertilizer usage. We propose that this hybrid plant modeling and data-driven approach for discovering new strategies to optimize crop yield may help address upcoming global food demands due to population expansion and climate change.