Digital technology is a means of interacting with and storing data. The difference between analog and digital technology is that analog technology translates data into various amplitude electric rhythms, whereas digital technology converts data into single amplitude electric rhythms. Information is translated into the binary system in digital technology, i.e., zero or one, with each bit representing two amplitudes.
In other words, when data is stored, transmitted, or forwarded in digital format, it is converted into numbers, or, at the most basic machine level, “zeroes and ones”. The term “digital technology” refers to any technology that uses some kind of processor to read this information, such as computers and applications that rely on computers, like the Internet, as well as other devices like video cameras, or mobile devices like phones, sensors, antennae, among many others.
Digital technologies cover a vast amount of innovations, from the automation of systems to the optimization of the decision-making process. This numerous group of technologies has revolutionized the way we do business. In the case of aquaculture, the wave that incorporates production with this kind of technology (along with some other innovations) is known as Aquaculture 4.0 or precision aquaculture. In this article, we will review the most promising digital technologies in aquaculture and their main current applications.
Automation consists in using technology to perform tasks with almost no need for human intervention. It can be implemented in any industry where repetitive tasks are carried out. In the case of aquaculture, there are several tasks susceptible to automation, being the most common feeding, water quality assessment, and counting. The inclusion of sensors, feeders, counters, and other similar machines helps to increase data flows, reduce labor costs, improve management and control of the production ponds. This also leads to improved biosecurity; since the number of people interacting directly with shrimp is reduced, the possibility of introducing diseases decreases as well.
Even though strictly automation is not necessarily a digital technology (although in most cases it is), it is one of the “mosts” to use other digital technologies due to the capacity to generate information and make the processes quick, efficient, and with smaller error. This group of technologies allows increasing production, both in area and intensity. Furthermore, since human error is reduced, productivity can be improved, and risks lessened.
Automatic feeders are already very advanced; when including the IoT that will be discussed in the next point, feeding can be done anywhere in the world and with optimal rations. For water quality, there are a vast number of sensors and integrated systems that allow us to gather data every minute (or less if necessary), letting us know different indicators, being the most used Dissolved Oxygen, temperature, pH, salinity, and alkalinity. This constant flow of information allows automating other equipment such as aerators or water pumps.
In automation, the leading research right now is the development of counters, which are especially difficult in shrimp production due to the turbidity of the water and the presence of noise. In this regard, three main strategies have been applied to counting: methods based on sensors, based on computer vision, and based on acoustic technology. Because of the difficulties mentioned above, the methods based on acoustic technology have been the ones that get better results in shrimp, being able to obtain high-quality images through sonar even with dark, turbid waters.
Usually, the automation of a farm requires a significant investment, first in equipment and then in maintenance and management, which results in a smaller but more specialized workforce. This aspect is the main bottleneck for implementing these technologies, especially in cheap labor countries (such as Latin America and Southeast Asia), where they opt for a more artisanal production rather than investing in automation.
The fact that most shrimp-producing countries belong to the developing countries group explains the reduced application of these technologies in shrimp as compared to salmon, which has more automation and digital technologies developed and applied industrially, and is produced in higher-income Countries (like Norway, the UK, Australia, or Canada).
Despite the apparent logic behind manual labor in shrimp, there is a lack of formal analysis for this strategy. The costs of implementing sensors and machinery should be contrasted, not only with the hiring of labor but also the increased efficiency and productivity, as well as the risk reduction and sustainability of the industry as a whole.
IoT and Cloud computing
IoT, which is short for the Internet of Things, consists basically in the connection of digital objects with the Internet. At the same time, cloud computing refers to the practice of employing a network of remote servers hosted on the Internet for storage. This sounds simple and not so techy, but it has enormous implications when combined with the rest of the digital technologies mentioned in this article. The IoT allows the farmer to be connected with the production at all times, making it possible to understand the situation better and make safer, informed decisions.
The application of the IoT to aquaculture is increasing at an exponential rate, with more applications in fish production. A salmon farmer can now see the fish he’s rearing, which are in sea cages far away from the coast, from anywhere in the world. This allows him to observe if there are any strange behaviors, which might indicate a disease, see if the salmon are eating well or if he is feeding too much, and keep predators at bay. This connection also permits observing and storing water quality indicator information, creating a production history, and obtaining more output information. Furthermore, suppose the farm is highly automated. In that case, it can be connected through the IoT and create a central management board, making all decisions centralized, reducing inefficiencies, redundancies, and, overall, optimizing production.
In the case of shrimp, this kind of technology is only starting to gain momentum, mainly in water quality assessment and control. But the future is promising when merged with other digital technologies, such as Artificial Intelligence and Big Data, which might help detect diseases before an outbreak, improve feeding and optimize management from a centralized office, allowing to produce in more remote areas.
Business Intelligence and big data
As stated in the first point of this list, automation of production allows generating an enormous amount of data in one production cycle. With the proposed digital technologies so far, that information would only serve to control and secure the quality of production in real-time, but the data generated would remain useless. This is what is happening in many shrimp farms where data is gathered manually and kept in physical logs and archived, or is rewritten in a spreadsheet and kept in a digital archive that, besides, is decentralized. It is precisely the lack of exploitation of the data generated that makes most farmers shy away from automation.
So the question arises, what can I do with all this data? The answers are multiple, being one of the most interesting for farmers to use this data to improve productivity and, in the end, increase their benefits. This can be achieved through the use of data science, in particular, with business intelligence.
Business intelligence (BI) refers to the set of strategies, applications, data, products, technologies, and technical architecture, which are focused on the administration and creation of knowledge about a system through the analysis of existing data in an organization or company. In the case of aquaculture, the use of bioeconomic modeling combined with big data allows generating recommendations for the optimization of the operations, making the farm more profitable and more sustainable using fewer resources. As stated in the definition, the use of BI depends hardly on the availability of big data sets; these data sets allow the modeler (or the software) to produce better parameters specific to the farm, which in turn produce better forecasts and improved recommendations. It’s important to remark that, in addition to a large quantity of data, its quality will also determine the accuracy of the implementation of BI.
Some of the most current applications of BI are the development of KPIs and their benchmarks, the forecast models to determine the best management strategies, the improvement of environmental efficiency, the development of growth strategies, the management of risk and negotiation of insurance premiums, and other investment analysis, such as the acquisition of new genetic lines, the effectiveness of new technologies, or the purchase of new infrastructure.
Artificial Intelligence, Machine Learning, and Deep Learning
In recent times, the increased computing power has made the development of complex and powerful algorithms that can “learn” from a data set and make decisions based on that data possible. We can find powerful tools such as machine learning, deep learning, and artificial intelligence among these computing techniques.
Previously, when talking about automation, we discussed the development and application of counters in aquaculture. The use of these data-driven tools has helped in the development of said counters. By training an algorithm with a large amount of known data, a software engineer can develop a program that learns from images, sound waves, reflection, or any known data that might help identify the number of shrimp in a pond. The more data the program has available, the better it gets predicting it.
Applications for these groups are immense, and they’re barely in their infancy stage. One exciting application being developed is feed optimization tools driven by the sound that shrimp do when eating. By implementing sound detectors in the pond and with known data, an algorithm can be trained to know how many shrimp are in the pond and how much they are eating.
Another interesting application is detecting and preventing disease outbreaks before they even happen by image recognition and water quality assessment. This has already been implemented in the appearance and treatment of sea lice in the Norwegian salmon industry, and it has started to be applied by Indian shrimp producers. Another application in this regard is diagnosing disease and the recommended treatment without recurring to a laboratory, reducing biosecurity costs.
Other applications are the screening of healthy seeds, water quality monitoring and prediction, and processing. It has been demonstrated that the use of AI in aquaculture can reduce wastage and decrease costs by up to 30%.
Geographic information systems (GIS)
Aquaculture has various space (and time) related issues, most of which can be associated with environmental and social aspects, such as mangrove area depletion, area expansion, site suitability, production impact, and livelihood and socio-economic impacts. To track and tackle these issues, the aquaculture stakeholders have looked into the use of GIS. Nonetheless, governments, research groups, and NGOs have promoted these methods more than the farmers themselves.
Currently, GIS technology is gaining attention, not only because of the numerous policy and sustainability applications possible but also as a tool to improve management, control, and productivity.
Some studies have been made regarding the possibility of using satellite imagery to monitor and gather water quality information. This would be a significant breakthrough since it would allow monitoring large areas with marginal cost as opposed to installing a substantial number of sensors in the ponds. This way, GIS information could be linked with machine learning to develop business intelligence protocols and management with a fraction of the cost associated with full farm automation.
Furthermore, the development of new low-orbit satellites increases images’ availability, resolution, and spectrum bands while reducing their costs. This opens a unique panorama for machine learning and AI as to how they interact with ponds since bands gather spectrums that capture information that is impossible (or incredibly expensive) to obtain with currently available technology.
Although we presented these digital technologies as separate tools, they work together to produce high-impact applications that improve shrimp farming most of the time.
Even though these technologies already exist and have been successfully applied in agriculture and salmon aquaculture. The acceptance, introduction, and use are still the main issues for their full implementation in shrimp aquaculture. The application of innovative contracts and proof of value are a couple of several ways of boosting the use of digital technologies in shrimp farming, making it possible to increase profitability for the farmers, reducing unit production costs, increasing production, and improving control, which in turn reduces risks and creates information, opening the panorama for more significant capital investors, cheaper credits and insurance primes, and further industry growth overall.