Towards better understanding of plant development through a core concept from artificial intelligence


Towards better understanding of plant development through a core concept from artificial intelligence

Published on
March 15, 2017

Today two ‘Wageningen-scientists’ published an insight review about the relation between external factors and the developmental and evolutionary processes in plants. For making this connection between ‘gene and Gaia’, they introduce the ‘perceptron’ concept, developed in the field of machine learning, into plant sciences and ecological trait analysis. Ben Scheres of Wageningen University & Research (WUR) and Wim van der Putten of the Netherlands Institute of Ecology & WUR conclude that considering diversity in plant responses to the environment as the adaptation of an information-processing ‘neural network’ opens up new avenues for studying the mechanisms underlying life-history strategies, trade-offs and evolution in plants.

Scheres and Van der Putten state that further research on the amazing flexibility plants in responding to their environment may be used for exploring novel approaches targeted at effectively enhancing the sustainability of future world food production.

Figure 2b

This is a graphic representation of the connectivity in plant growth-regulatory networks as presented by Scheres & Van der Putten. The intrinsic developmental program in plants sets up spatially restricted domains of growth-factor signalling and their response systems (upper layer of the network). Polar auxin transport (PAT) is shown as an example. Cross-talk between growth factors (brassinosteroids (BR), gibberellic acid (GA), auxin (AUX), cytokinins (CK) and ethylene(ET)) occurs through signal-transduction pathways, which form ‘hidden’ layers that integrate information by changing their activity in response to inputs.

Ultimately, the hidden layers control transcription in the output layer (bottom row). Nodes in the output layer represent genes with promoters that integrate weighted inputs from the previous layer. A single output node in the drawing may represent several genes, the encoded proteins of which control a developmental process.

By connecting this network to external signal such as light, nutrients and signals from the plant immune system, the plant can respond very specifically to the external world.

Feedback between different information-processing nodes is indicated by red lines.