Very recently, Dr Lee Hickey of the University of Queensland told a short story about a farmer in the future being alerted to a disease outbreak on his farm via his phone. He swiftly commands one of his drones to spray an RNAi specific to the disease on the infected field and averts the crises. Instead of worrying about the problem and the potential effect on his crop’s yield for the year, he goes to the footy with mates without a worry.
The idea of remotely sensing and specifically addressing a crop under stress is a current area of technology under development. There are some commercial sensors available to detect a small array of stressors and the use of drones in agriculture is increasing, but the ability to monitor and address a number of potential in-crop problems at the same time is a technology still confined to a distant world in an Isaac Asimov novel.
But that doesn’t mean scientists aren’t researching how the technology might work.
A recent study in the journal Agriculture took three commercial sensors, two being spectrometers and one a fluorometer. The spectrometers can use a variety of indices to give an output which relates the amount of light hitting a plant and the amount of light being reflected to, for example, how green the plant is. Fluorescence sensors measure the fluorescence of the cells of the plant and can adjust to measure such spectra as red fluorescence, blue green fluorescence and infra-red fluorescence. From these measurements certain conclusions can be made about the health of the crop being monitored.
Although the sensors mentioned above are normally used to monitor an individual stress such a water deficiency, nitrogen deficiency, fungal infection or weed competition, this paper used a combination of stress of factors to determine whether a single band or index could be used to detect and identify a single stress whilst one or more other stressors were present.
The experiment was set up so that every combination of the four types of stress and a control was tested using each index available on each device. 12 days after planting, the water stress testing commenced with water reduced to 30% of the pot water holding capacity while non-stressed plants were watered at 70% of the water holding capacity. For nitrogen stressed plants no nitrogen fertiliser was used throughout the experiment while the non-stressed wheat received fertiliser at predetermined amounts on predetermined days. Weed plants and fungal infections were added at the same stage of the experiment.
Table from article. Combinations of stress treatments.
Each treatment combination was measured every third day after two leaves had appeared on the spring wheat that was being tested. For each index to be measured, five readings were taken on each reading day, and the five readings averaged.
Each stress resulted in a significant effect on the growth of the spring wheat either by the amount of biomass or the ratio of root to shoot growth. Many indices provided significantly different spectral ratios and fluorescence levels between control plants and plants tested with one type of stress. The tables of average values for each index tested for each device and whether the difference between stressed and non-stressed values were statistically significant are provided below.
For a given index, combining the average values of more than one stress made it more difficult to ascertain whether a plant was stressed and, if so, which stress was being applied.
Without pulling the entirety of the article here, an example of this difficulty is given. Where two different types of stress resulted in an increase of the index value for one stress and a decrease in the index value for the other stress, determining whether one or the other stress is present is easy – if the value goes down compared to the control, then the individual stressor that causes the index to be lower is present, and if the value is higher than in the control then obviously the stressor that increases the value is present. However, if both stressors are present then the value given will be somewhere in between the two extremes and can be difficult to distinguish from the control. A graph of the values of the FLAV index from the Multiplex sensor reading nitrogen deficiency and water shortage is given as an example of this difficulty.
From Figure 7 of article (“-” = not stressed, “+” = stressed). Using the FLAV index for water and nitrogen monitoring, the unstressed plant gives the lowest reading, the water stress only treatment gives the highest reading, the nitrogen stress only treatment gives a reading only slightly higher than the unstressed treatment, and the treatment with both types of stress gives a reading roughly half way in between the unstressed and the water stress only treatment.
The difficulty caused by this is that, pretending you were the farmer in Dr Hickey’s story, it may be difficult for the app on his phone to differentiate between a nitrogen stressed crop, an unstressed crop and a crop with both nitrogen and water stress. The lack of precise diagnosis results in the lack of precise management.
What was also clear from the data obtained was that an index value for an unstressed plant at one particular stage of development can be identical with the value for a stressed plant at a different stage of development. The result is an added complexity in differentiating between a stressed and unstressed crop as a function of growing time compared to a stress that can be differentiated from a control plant at the same stage of growth.
The experiment shows that there are a number of current indexing methods of differentiating between a plant faced with a particular stress to an unstressed plant. However, for some indices the stressed and unstressed plants need to be at the same growth stage in order for the results to be of use. Further, while some indices are able to show the presence of one type of stress when other types of stress are also present, the interplay between the effect of different stressors can make detecting a stress or identifying whether one ore more particular stressors are present difficult.
For the time being, developing a set of standard readings over a range of indices for unstressed and differently stressed species at different growth stages is required. The standards can be incorporated into sensors that may use either complex algorithms to determine whether a plant is stressed and, based on the deviation of the reading to known standards for a range of types of stress, evaluate the likely type or types of stress present. Alternatively, a sensor using a variety of indicies which can each individually distinguish a different type of stress and deviation from a set standard may result in more accurate and dynamic sensing system.
We are still quite a way from Dr Hickey’s vision, but the technology and the methods to distinguish different types of crop stress already exists. Its not hard to envision a breakthrough piece of technology becoming available soon, and harder not to imagine that this type of technology will soon be incorporated into automated precision agriculture such as the Farmbot and indoor agricultural systems.