Haoyan Yang, Lingju Zeng, Xiangjian Gou, Yang Shao, Xianlong Song, Shuang Zhang, Xiaoyan Qian, Zhenwei Zhang, Xiao Wang, Haitao Li, Bo Peng, Juliet Nkiruka Anyanwu, Zihao Zheng, Linqian Han, Ling Zhou, Mustafa Bulut, Peng Song, Wanneng Yang, Yingjie Xiao, Wenqiang Li, Mingqiu Dai, Fazhan Qiu, Jian Zhang, Baobao Wang, Alisdair R Fernie, Haiyang Wang, Han Zhao, Xianran Li, Jianbing Yan, Tingting Guo
A convolutional neural network (CNN) trained on 25,080 maize images achieved 96.7% accuracy in distinguishing plant responses to low- and high-N conditions, revealing previously unrecognized phenotypic variation associated with NUE. Deep phenotypes showed greater phenotypic variation and higher heritability compared with conventional phenotypes, enabling the identification of 523 significant loci compared with 21 for conventional phenotypes. Functionally characterized Liguleless2 (LG2), a basic-leucine zipper (bZIP) transcription factor, demonstrated that lg2 mutants exhibit enhanced root architecture and increased N uptake efficiency, with field trials supporting the AI findings.
Improving nitrogen use efficiency (NUE) is essential for sustainable agriculture, yet conventionally measured plant characteristics have limited value as NUE proxies. Here we show that artificial intelligence (AI) can uncover previously unrecognized phenotypic variation associated with NUE, revealing genetic variation that is largely missed by conventional phenotypes. We trained a convolutional neural network (CNN) on 25,080 maize images to learn features that distinguish how plants respond to low- and high-N conditions, achieving 96.7% accuracy. The learned features were defined as deep phenotypes. Compared with conventional phenotypes, deep phenotypes showed greater phenotypic variation and higher heritability, enabling the identification of 523 significant loci compared with 21 for conventional phenotypes. We next investigated candidate genes underlying these loci and used these findings to interpret the learned features. Lower CNN layers primarily reflected visual patterns overlapping with conventional phenotypes, whereas deeper layers encoded additional features associated with N-responsive genetic variation. To validate candidate genes identified by the AI framework, we functionally characterized Liguleless2 (LG2), a basic-leucine zipper (bZIP) transcription factor, and demonstrated that lg2 mutants exhibit enhanced root architecture and increased N uptake efficiency. Field trials of 200 hybrids across diverse N environments further supported the AI findings, with each beneficial allele increasing ear weight by an average of 18 g per plot under low-N conditions. These results show how integrating AI and biology can uncover biologically relevant variation underlying complex traits such as NUE and enhance the interpretability of AI models.