Integrating nutrient demand models and AI-based sensors with precision-dosing rigs to improve resource use and productivity, and reduce waste and emissions in commercial raspberry production.

Soft fruit is an exciting product area with excellent growth potential. Although UK soft production is growing by ca. 8%/year, **demand for** berries by UK consumers still exceeds supply. Continued growth is needed to displace often inferior imports, but this must be achieved on a sustainable basis through efficient utilisation of valuable resources (primarily water and inorganic fertilisers) and minimal environmental impact.

Soft fruit growers know that a sub-optimal supply of macro- and micro-nutrients will limit marketable yields and berry quality, but most guidelines on fertiliser inputs are hopelessly outdated. These formulations are often adjusted based on anecdotal observations by growers and agronomists, but there is little scientific basis to these amendments and many unneeded macro- and micro-nutrients accumulate in the substrate. Growers then apply irrigation flushing events to remove these harmful so-called "ballast ions" which wastes water, can result in lowered berry firmness, flavour and shelf-life, and poses a risk to local groundwater quality.

Excessive N inputs often result in elevated emissions of N2O as a result of denitrification, and N2O emissions account for ca.44% (global warming potential \[GWP\] basis) of the total agriculture-related GHG emissions. CO2 has a GWP value of 1 while N2O has a value of 298, making the latter a more potent GHG. Reducing N inputs in agriculture and horticulture by more closely matching demand with supply should help to reduce N2O emissions, but this is a risky strategy if guidelines and monitoring sensors are not available.

Our nutrient demand modelling work in IUK 102124 showed that N input to substrate-grown raspberry could be reduced by 32% without affecting marketable yields and berry quality, and overall water and fertiliser demand was lowered by 20% due to a reduction in plant biomass (less luxuriant growth). In a follow-up project IUK 102640, we have developed a prototype AI-based nitrogen / phosphorous / potassium (NPK) real-time sensor that growers can use to determine NPK availabilities in coir to inform their fertigation decision making. Initial testing has revealed a number of technical issues with the prototype system which need to be resolved before this strategy can be developed further. Work will be carried out with the sensor manufacturer to address sensor limitations and the output of this work will be evaluated under commercial conditions in 2022. We will also determine an effective exploitation strategy for the sensors as a stand-alone hand-held system and evaluate the integration of the sensors with irrigation controllers and new technologies to improve/automate crop nutrient management.

This project will demonstrate the potential to combine a new variety-specific N demand model with a prototype sensor system that estimates NPK coir availabilities in real time. This approach will then be integrated into the NetBeat(tm) platform and combined with other technologies to deliver an automated control system. The SmartNutrigation system will maintain coir NPK availabilities within a narrow optimum range during each developmental stage using outputs from nutrient demand models and real-time feedback from the NPK sensors thereby maximising sustainability. It will deliver improvements in crop quality and productivity and improve nutrient and water-use efficiency, thereby creating a more sustainable production system.