TY - JOUR
AU - Probst, Wolfgang, Nikolaus
AB - Abstract The ubiquitous spread of digital networks has created techniques which can organize, store, and analyse large data volumes in an automized and self-administered manner in real time. These technologies will have profound impacts on policy, administration, economy, trade, society, and science. This article sketches how three digital data technologies, namely the blockchain, data mining, and artificial intelligence could impact commercial fisheries including producers, wholesalers, retailers, consumers, management authorities, and scientist. Each of these three technologies is currently experiencing an enormous boost in technological development and real-world implementation and is predicted to increasingly affect many aspects of fisheries and seafood trade. As any economic sector acting on global scales, fishing and seafood production are often challenged with a lack of trust along various steps of the production process and supply chain. Consumers are often not well informed on the origin and production methods of their product, management authorities can only partly control fishing and trading activities and producers can be challenged by low market prices and competition with peers. The emerging data technologies can improve the trust among agents within the fisheries sector by increasing transparency and availability of information from net to plate. Introduction Smartphones influence our lives through their multi-purpose versatility. They allow us to communicate, orientate, inform, educate, and shop at almost any location at any time. Several users touch their smartphone more than hundred times per day, and excessive smartphone use can even lead to mental addictions and psychological disorders (Kwon et al., 2013; Haug et al., 2015; Samaha and Hawi, 2016). Our reliance on smartphones indicates how ubiquitous the use of digital networking technologies have become in everyday life. Currently, a large proportion of digital innovation is focusing on gaining extra information out of the large volumes of exchanged data. The economic aim of this data gathering is to turn the inherent information into added value by gaining better insight into the behaviour of consumers and their demands, but also to make production and trading processes more efficient by monitoring the flow of products, goods, and services. Digital integration is advancing in many economic sectors and thus it is only reasonable to assume that it will also affect fisheries production and associated economic branches. This article sketches potential impacts of emerging internet-based data techniques, i.e. the blockchain, real-time data mining, and artificial intelligence (AI), and shows how these techniques can influence the way fisheries are operated and managed and fish products are traded and consumed. The article will focus on these three technologies, because they are based on gathering, organizing, recording, processing, and analysing large volumes of data and are expected to drive major technological, economic, and social developments in the future (Li et al., 2015; Swan, 2015; Tapscott and Tapscott, 2016; Cath et al., 2018). While data mining and AI are not new disciplines by themselves, they have gained renewed interest by big-data applications and digital innovations, such as autonomous driving or image recognition (Rajamaran, 2016; Cath et al., 2018). Fisheries: a global and elusive enterprise Fish and seafood products are harvested and traded worldwide (FAO, 2016). For many of these products, it is a long way from net to plate, as their supply chains cross multiple continents, national borders, trade zones, and jurisdictions. For consumers and national controlling agencies, it is therefore not easy to recognize the true origin of imported seafood products. However, even products of regional fisheries can be associated with knowledge gaps on catches, by-catches, and environmental impacts (Hall, 1996; Lewison et al., 2004; Kaiser et al., 2006; Benoit and Allard, 2009; Herr et al., 2009). Consumers, producers, and management authorities are therefore faced with the challenge to place trust into producers and the members of the supply chain, while knowledge on the exact behaviour of the other participants is limited (Mosler, 1993). In many commercial fisheries, in which the fishing companies are referred to as “producers”, only a proportion of the full catch (or capture) is retained and landed (Hall, 1996). At harbour, the landed catch is reported to wholesellers and management authorities before being traded and processed to be finally sold to the consumers. What really happens on board of fishing vessels, i.e. which species in which quantities are really caught, remains elusive to management authorities for the majority of fishing operations (Benoit and Allard, 2009; Edgar et al., 2016). Along the supply chain, various mechanism can operate to control the compliance of the supply chain members to regional and national jurisdictions. At sea, the national coast guard, observers-at-sea, or video-monitoring systems can inspect fishing practices or record the catch (Ulleweit et al., 2010; Kindt-Larsen et al., 2011; Haskell et al., 2014), but the frequency of these control mechanisms is often negligible compared to the frequency of total fishing operations. On land, the most important control measure is the declaration of the landings to national authorities to allow the comparison against the allocated annual catch quota. Depending on the world region of landing, this mechanism covers a smaller or larger proportion of the landings, as not all landed species are quota restricted. Also at land, the production and processing of seafood products are controlled by governmental or self-committed industry institutions (consumer or eco-labels) (Gulbrandsen, 2009). During all steps of the supply chain, the mechanisms of control can be cheated, leading to unreported and/or illegal catches (Helyar et al., 2014). During the trading and processing chain, landings can also be re-declared into another, more valuable species or being caught from a different origin (stock or catch area). Increasing the traceability of fish and seafood products for consumers and management authorities has therefore become a designated aim of retailers and policy makers (Schröder, 2008). Data technologies on the rise The lack of transparency in fisheries production and trading processes has led to a trust crisis by consumers into management authorities and the industry (Jacquet and Pauly, 2007; Helyar et al., 2014). Emerging data technologies may help to overcome some aspects of this crisis by improving the transparency for controlling agents such as end-consumers, NGOs, and management authorities. Blockchain and smart contracts Blockchains have become prominent through the boom of crypto currencies, for which blockchains provide the technical foundation. Since then blockchains have been modified in multiple way leading to an ever increasing number of blockchain projects and crypto currency tokens. But blockchains are not only used for crypto-currencies. In fact, many companies and independent institutions are currently looking for possible implementations of blockchains into their operations (Swan, 2015; Xu et al., 2016, see also the IBM website for examples of blockchain applications). In October 2008, an anonymous author with the pseudonym Satoshi Nakamoto released a whitepaper for a non-institutional (i.e. non-governmental) currency system called Bitcoin. Bitcoin is based on a blockchain that records financial transactions in a decentralized database. Thereby, Bitcoin was originally perceived by a community of enthusiast as an alternative and independent financial system, which could overcome institutional structures which lead to the financial crisis of 2008. The blockchain database, generally referred to as distributed ledger, adds new entries chronologically within a blockchain network (Swan, 2015, Figure 1). The entries are combined in blocks, which are linked by checksums (numbers to validate the integrity of data) to the previous block (Christidis and Devetsikiotis, 2016). This linkage as well as the decentralized and synchronized storage on all participating nodes of the blockchain network are supposed to make blockchains invulnerable to subsequent data manipulation and hacking. At the moment, blockchains find their widest application in storing transactions of crypto-currencies such as Bitcoin or Etherium, but they can contain any sort of information, such as text, documents, images, or music. Figure 1. View largeDownload slide A simplified representation of blockchains. (a) Transactions are coded by hash values, which in turn are combined into blocks within the block chain. Single blocks are connected via the hash of the previous block and the hash values of new transactions. (b) The block chain is shared among the participants of the network, with each participant holding an identical copy of the blockchain. New blocks must by validated by a critical number of network participants to be added to the blockchain. When this happens, the blockchain version of all network members will be updated. (c) Smart contracts are implemented within the blockchain, i.e. are blocks that are referred to by contracting parties (here partners a and b). The smart contract validates the requested action and if approved, executes its content (e.g. a conditional transaction coded with the contract). The executed contract will be amended to the blockchain when approved by the network. Figure 1. View largeDownload slide A simplified representation of blockchains. (a) Transactions are coded by hash values, which in turn are combined into blocks within the block chain. Single blocks are connected via the hash of the previous block and the hash values of new transactions. (b) The block chain is shared among the participants of the network, with each participant holding an identical copy of the blockchain. New blocks must by validated by a critical number of network participants to be added to the blockchain. When this happens, the blockchain version of all network members will be updated. (c) Smart contracts are implemented within the blockchain, i.e. are blocks that are referred to by contracting parties (here partners a and b). The smart contract validates the requested action and if approved, executes its content (e.g. a conditional transaction coded with the contract). The executed contract will be amended to the blockchain when approved by the network. Blockchains solve the problem that electronic transactions between two partners have to be mediated by a third authority, e.g. a bank, payment service, or seller (Tapscott and Tapscott, 2016). Instead, the transactions are verified by the blockchain network, which is self-organized and cannot be manipulated afterward. Therefore, blockchains are ideal in situations in which the trust of network participants into any mediating/controlling party is limited, i.e. that centralized administration of transactions is not trusted or technically not feasible. The latter case may arise when for example supply chains of sea food products cross the limits of multiple responsible controlling and surveillance authorities, e.g. by imports from Asia into the European Union or North American market. Or if any sort of regulating authority is totally absent, e.g. in remote, artisanal fisheries with unregistered fleets. Smart contracts are self-executing scripts included into blockchains that automize predefined operations, e.g. trading rules for crypto currencies or other assets (Christidis and Devetsikiotis, 2016). Smart contracts thereby extend the functionality of blockchains to allow for more complicated operations than just simple transfer of assets, e.g. conditional trades including “if-then” functions and other conditional rules. Further functionalities are constantly added to blockchain platforms and their technological development is far from finished. Current efforts are focusing on increasing the speed of blockchain transactions while reducing the energy requirements to maintain the network. Data mining and big data The global use of digital devices generating increasing amounts of data is growing fast. In 2015, presumably about 8 zettabytes of e-mails, blogs, social media posts, images, and videos were created (Rajamaran, 2016). Accordingly, architectures and tools for storing, exchanging, handling, and analysing large volumes of data (“big data”) have been developed (Zakir et al., 2015). The process of analysing big data is commonly referred to as data mining (Kantardzic, 2011) and includes methods to describe patterns within the data and to predict events from these data. Data mining itself is not a new branch of information technology, but it has been boosted with the demand for real-time analysis of large data volumes in web-based applications and digitized industries (Zakir et al., 2015). Typical data mining tasks are the identification of outliers, classification, and clustering of data, regression analysis as well as condensing data into summaries and overviews. While many analysis in natural sciences traditionally include some form of data mining, in the context of this article data mining techniques are commonly associated with big data gathered for non-scientific purposes (Walker, 2014). Big data are usually described by volume, commonly ranging in the terra- to petabyte domain, but big data can also be classified by variety and velocity (Russom, 2011; Rajamaran, 2016). Variety refers to the type of data available, e.g. structured, semi-structured, or unstructured, depending on the sources the data is coming from and which data sources are combined. Classical sources for big-data analysis are data warehouses, in which several data bases of different format and content are combined. Velocity refers to the frequency of data streams, i.e. whether data are updated or added in real-time, near real-time, or in batches. Especially real-time data are most likely to have the strongest potential for innovative applications, but are also the most demanding with regards to storage, processing, and analysing. Artificial intelligence AI is a branch of computational sciences which deals with the ability of machines, i.e. computers to achieve goals by learning based on previous experience (Russell and Norvig, 2010). AI is therefore often used interchangeably with the term “machine learning”, even though AI is a more generic concept, of which machine learning is only one aspect. The ultimate goal of machine learning algorithms is to come to automized decisions in non-determined situations resembling human cognitive abilities, such as recognizing objects in images or translating sentences from one language into another. AI finds application in speech and image recognition (Kantardzic, 2011; Ghahramani, 2015), crime prediction (Shapiro, 2017), or autonomous driving of vehicles (Urmson et al., 2008). In the wake of big-data applications, machine learning algorithms have become very popular, as they are able to learn as they are trained on existing data (Hastie et al., 2009; Tayal et al., 2014; Ghahramani, 2015; Rajamaran, 2016; Shapiro, 2017). The major asset of machine learning algorithms is their ability to learn from the training data and to incorporate new data into the learning process as the data flow into the database. Machine learning algorithms are often categorized into supervised and unsupervised learning algorithms (Hastie et al., 2009). Supervised learning are based on some sort of model, in which data are separated into input and output features, i.e. into predictor and response variables (Kotsiantis, 2007). Some popular supervised machine learning algorithms are decision trees, support vector machines, Bayesian networks, boosted regression trees, random forests, and artificial neural networks. They can be applied for medical applications such as cancer prognosis and prediction (Kourou et al., 2015) or the analysis of social media content (Ruths and Pfeffer, 2014). Unsupervised learning methods analyses the data structure without distinguishing between input and output variables, with many unsupervised algorithms working around clustering and discriminating data cases according to their features (Hastie et al., 2009). A famous example of a non-supervised machine learning algorithm is the Google PageRank algorithm for web searches in the World Wide Web. How can Blockchain & Co. improve the trust fisheries? Blockchain One of the few documented examples to implement blockchains in fisheries is the attempt to support the traceability of tuna around Fiji and other South Pacific islands (Visser and Hanich, 2017). To combat illegal, unreported, and unregulated (IUU) fishing, participating fisheries attach radiofrequency identification (RDFI) or quick-response (QR) code tags to each fish immediately after its catch. These tags are then registered automatically at various stations of the processing (i.e. supply) chain and each registration is fed into the blockchain. According to the example of the South Pacific tuna fishery, a UK-based company called “Provenance” implements blockchains for agricultural, forestry, and fishing products including faire-trade and organic consumer labels (www.provenance.org). These application exemplify several advantages of blockchains, i.e. the registration and processing of the product is automized and the data in the blockchain is supposed to be tamper proof as it should not be modifiable by a single member of the supply chain (Pfreundt, 2018, but see section on problems on caveats for safety issues with blockchains). Furthermore, the blockchain does not have to be organized by a single controlling agency (non-governmental or governmental institution), but is self-organizing across jurisdictional borders and institutional responsibilities. And finally, the blockchain transactions are transparent, allowing all participants of the market, including the consumer, to track the origin of the fish. This does not mean that tracking of supply chains by blockchains are invulnerable to fraud and cheating. In the example of the South Pacific tuna fishery, the most crucial step within the supply chain is the correct labelling of individual fish. Mislabelling can still happen and with criminal energy, fishers, traders, ex-, and importers still may find ways to land and sell ill-declared fish. But the transactions registered within the blockchain will be unchangeable and hence cheating of a single party within the supply chain should become much more difficult. Blockchains may help to monitor landed and traded fish along the supply chain much better than current, centralized databases, which are often restricted to national or regional jurisdictions. Data within blockchains could become a valuable source for enforcement agencies, fisher, traders, consumers, and scientists (Pauly and Zeller, 2003; Sumaila et al., 2007; Mora et al., 2009) to analyse catch and landing volumes as well as revenues of producers, processers, and traders. At the moment, these data have to be actively gathered by governmental agencies requiring significant funding and manpower (Stransky et al., 2008; Dörner et al., 2018). If landed fish and seafood would automatically register to a blockchain, landing volumes of a species could be counter checked with global retail volumes to identify and track discrepancies. Even if this sort of tracking would not work on a mandatory basis, as many participants in the fishing industry may lack the capacity or the will to participate in a fully electronical processing system, the participation could occur on a voluntary basis within consumer-labelling schemes, e.g. organic, sustainability or fair-trade labels (Gulbrandsen, 2009; Swan, 2015; Visser and Hanich, 2017). In such labelling schemes, landing shares as tonnes of biomass could be purchased by fishing companies to produce and trade these shares within the blockchain of the label. This would turn catch quotas into a blockchain asset similar to a crypto currency. Alternatively, governments (at least in developed countries) could combine the allocation of catch quotas to fisheries organizations with the obligation to apply a blockchain-based tracking system. Catches without a share in the blockchain could thus not be landed and sold, thereby becoming illegal. When catch quotas are traded as blockchain assets, any quota share (e.g. 1.000 t of herring) would then be attributed with a volatile trade value, similar to stocks in a stock exchange. If the actual value of the fish within the quota was higher than the trade value of the quota, fishermen could decide to actually fish the quota out and the asset would be annulated from the market. If the trade value was higher than the price for the fish, fishermen could decide to keep or sell the quota as financial asset. This system would provide interesting options not only for fishing companies, but also for non-governmental conservation organizations (NGO) or management authorities to implement buy-out schemes. The trade value of quota shares could rise if the annual catch quota is low, providing financial compensation to fishermen. Finally, blockchains on landing trades may hold an advantage to fishermen themselves. If they would have access to all trades that have occurred at the port in their vicinity, they could choose the trader or port, which pays the best prices for their catch. Blockchains could also be used to trade catch quotas between fishing vessels, for example in fisheries with tradable catch quotas (Branch, 2009). Participating fishing vessels and management authorities would be informed on the ongoing trades in real time and also across national jurisdictions. Data mining and artificial intelligence Data mining and AI are tightly linked, as many AI algorithms require large datasets and some data mining techniques are in turn based on AI (Kotsiantis, 2007; Hastie et al., 2009; Kantardzic, 2011). A prominent example of combined data mining and AI is predictive policing (Tayal et al., 2014; Shapiro, 2017). Predictive policing uses georeferenced data (e.g. unemployment rate, population density, financial income of residents, etc.) to anticipate in which areas specific crimes such as burglary, mugging, or murder may occur at significantly high rates (Pearsall, 2013). Alternatively, some predictive policing software can estimate how likely previously convicted persons may commit another crime (Shapiro, 2017). Similar to predicting robberies and muggins in a city, it is possible to develop decision support tools based on AI and data mining for law enforcement agencies at sea. An example is a decision support tool for the US Coast Guard in the Gulf of Mexico to choose locations for patrolling against illegal fishing (Haskell et al., 2014). This tool is based on a game-theoretic model which predicts the response of illegal fishermen to the patrolling scheme of the US Coast Guard in the Gulf of Mexico. While this example represents a situation with limited data availability on fishing activities, it is easy to envision even more powerful decision support tools in fully industrialized fishing fleets equipped satellite tracking devices and electronic logbooks. In these fisheries, reported catches and locations of fishing operations can be transmitted in near real time and analysed by data mining algorithms on land-based servers. If these algorithms detect significant deviations from the common pattern recorded for the according area, gear type, and season, patrols could be send out to inspect these vessels at sea. This kind of system would be most efficient in a fisheries in which the majority of fishers comply with the management rules, as the data created by the “black sheep” would stand out from the majority of logbook records and thus should be easy to identify by data mining algorithms. Data mining algorithms could be also used to improve the implementation of spatial real-time closure (RTC) in fisheries where high amounts of unwanted by-catch are observed. In 2009 the Scottish Government implemented a RTC scheme on areas of high catches of juvenile Atlantic cod Gadus morhua to reduce mortality and discarding (Holmes et al., 2009; Needle and Catarino, 2011). In the RTC scheme, data from on-board observers and satellite-based ship tracking (VMS) are used to determine areas of high catches. However, data can only be analysed after the reporting of landings at harbour and thus the designation of closure sites is associated with a temporal delay. To improve the implementation of RTC, catch data from e-logs could be transmitted to land-based servers, where they could be analysed by data mining algorithms even sooner. A prominent application of AI algorithms is the field of image recognition. In fisheries, AI-based image recognition could improve electronic observing-at-sea programmes based on video monitoring. At the moment, the recognition of species and catch volumes from electronic monitoring is commonly done manually, requiring human observers to sight video footage and image stills (Kindt-Larsen et al., 2011; van Helmond et al., 2014). Consequently, only small proportions of the recorded data are actually analysed. The automized recognition of species, length, and catch volume would allow to analyse more data at lower costs providing a more complete picture of the total removals of commercial fishing fleets. Some studies which have implemented AI-based image recognition to identify fish species in aquacultures (Hu et al., 2012) and in catch fisheries (Storbeck and Daan, 2001), achieved classification rates of >90%. Image recognition is also increasingly used to identify vessels at sea (Kanjir et al., 2018). Optical remote sensing images can be combined with satellite or land-based geolocation techniques, such as vessel-monitoring system (VMS) or automatic identification system (AIS) to identify ship size and activity type. Though the satellite-based classification of vessels at sea is still faced with challenges by cloud cover, solar light input angle and overflight frequency, refined algorithms based on machine learning may help to overcome some of these shortcomings to detect unreported and illegally fishing vessels. Even if this data may not become available in real-time, similar to radar controls it may be used as evidence for fining and sanction schemes. Apart from image recognition, AI may have many potential applications in fisheries and trading of seafood products. At the moment, implemented examples are scarce, but AI may find fruitful application in any situation in which large data volumes need to be analysed and categorised. Due to the versatility of AI and its potential applications, this article does not intend to provide a comprehensive overview on AI applications in fisheries, but rather intends to spark the creativity of software developers and decision makers in economic and management institutions. A word on the “internet of things” Internet of things (IoT) refers to the automized communication of electronic devices (Christidis and Devetsikiotis, 2016). Its development is tightly linked to the equipment of devices with sensors and communication electronics such as Bluetooth or wireless network adapters. Recent IoT applications are smart home products that can regulate lighting, entertainment systems, heating and locking of doors and windows. There are also many applications in agriculture such as satellite-based harvesting and planting machines, automized irrigation systems connecting soil sensors with pumps and water gates or radiofrequency identifiers (RFID) for tracking free-ranging cattle (Dlodlo and Kalezhi, 2015). Similarly, RFID have been attached to Tilapia reared in Chinese aquaculture to follow individual fish through the production and trading along the supply chain (Bo et al., 2012). This is very similar to the blockchain tuna example described above, the only difference being that the Tilapia a tracked without a blockchain. Without any doubt IoT applications will find many opportunities within fisheries productions and trades, but they are not in the focus of this review. While IoT and data applications are tightly linked, the former is very much associated with hardware developments, i.e. the application of sensors in previously not measured systems. It can easily be imaged that IoT can play a big role in processing seafood on-board of fishing vessels, e.g. by improving automized grading and gutting. Also a more direct communication between consumers and producers (fishermen) would be thinkable to enhance direct marketing of fresh caught fish, leading to better prices and short supply chains. However, these are just two of many possible IoT applications in fisheries, reviewing them all is beyond the scope of this article. Nonetheless it should be noted that several blockchains techniques are explicitly developed to accommodate requirements of IoT applications. A well-known example is the crypto currency IOTA, which the developers foresee as a native currency for financial transactions between autonomous devices (see www.iota.org). Meta-view on potential applications of blockchain, data mining, and AI Looking at Table 1, data mining and AI appear to be especially useful tools to monitor and control fishing vessels/companies by helping to gain knowledge on catches and to ensure compliance to management rules at sea. Thereby both techniques are acting mainly on the first link of the supply chain, the producer, i.e. the fishing vessel, whereas blockchains and smart contracts are useful to ensure transparency along the supply chain. It should be noted, however, that the implementation of each technology is associated with caveats. Table 1. Examples of potential applications of blockchains, data mining, and AI in economics and management of fisheries. Technology Task Strengths and opportunities Weaknesses and threats Blockchain Tracking of the supply chain of fish and seafood products by consumers and/or retailers Increase consumer trust and engagement in ecolabels and sustainability campaigns Not fraud proof as mislabelling/mal-codification still may occur Blockchain and smart contracts Trading of quotas/shares among fishing companies and between management authorities and fishing companies Transparent and self-organized trading of catch shares/quotas May not comply with management specificities Blockchain and data mining Tracing and verifying reported landings by management authorities Adds transparency to fisheries data Facilitated enforcement of fisheries regulations Allows to direct inspection efforts to critical cases Difficult to implement and demanding on IT resources Algorithms may lead to wrong conclusions Sell at sea of catch by fishing companies to wholesalers factories, retailers Improved trading opportunities for producers (fishers) May incentivise the landing in countries with less strict management obligations Data mining Designation of RTC Improved speed in closure designation Requires near real-time transmission of catch information Data mining and AI Predictive control against IUUa, decision support for patrols Improved allocation of inspection efforts, higher incentive for fishers to comply with laws and management rules Requires near real-time transmission of catch information, which can still be flawed AI Catch recognition in on-board video footage Improved processing of catch, better data collection for management authorities Difficult to implement technically, may not be error or tamper-proof, high costs for companies operating fishing vessels Vessel identification of optical remote sensing images Improved knowledge on vessel location and activity May have a deterrent effect on illegal fishing vessels Spatial and temporal coverage may have significant gaps Error rate in vessel identification may be high Technology Task Strengths and opportunities Weaknesses and threats Blockchain Tracking of the supply chain of fish and seafood products by consumers and/or retailers Increase consumer trust and engagement in ecolabels and sustainability campaigns Not fraud proof as mislabelling/mal-codification still may occur Blockchain and smart contracts Trading of quotas/shares among fishing companies and between management authorities and fishing companies Transparent and self-organized trading of catch shares/quotas May not comply with management specificities Blockchain and data mining Tracing and verifying reported landings by management authorities Adds transparency to fisheries data Facilitated enforcement of fisheries regulations Allows to direct inspection efforts to critical cases Difficult to implement and demanding on IT resources Algorithms may lead to wrong conclusions Sell at sea of catch by fishing companies to wholesalers factories, retailers Improved trading opportunities for producers (fishers) May incentivise the landing in countries with less strict management obligations Data mining Designation of RTC Improved speed in closure designation Requires near real-time transmission of catch information Data mining and AI Predictive control against IUUa, decision support for patrols Improved allocation of inspection efforts, higher incentive for fishers to comply with laws and management rules Requires near real-time transmission of catch information, which can still be flawed AI Catch recognition in on-board video footage Improved processing of catch, better data collection for management authorities Difficult to implement technically, may not be error or tamper-proof, high costs for companies operating fishing vessels Vessel identification of optical remote sensing images Improved knowledge on vessel location and activity May have a deterrent effect on illegal fishing vessels Spatial and temporal coverage may have significant gaps Error rate in vessel identification may be high Note: This list is not intended to be comprehensive, but represents some potential applications together with associated strengths, weaknesses, opportunities, and threats. a Illegal, unreported, or unregulated. Table 1. Examples of potential applications of blockchains, data mining, and AI in economics and management of fisheries. Technology Task Strengths and opportunities Weaknesses and threats Blockchain Tracking of the supply chain of fish and seafood products by consumers and/or retailers Increase consumer trust and engagement in ecolabels and sustainability campaigns Not fraud proof as mislabelling/mal-codification still may occur Blockchain and smart contracts Trading of quotas/shares among fishing companies and between management authorities and fishing companies Transparent and self-organized trading of catch shares/quotas May not comply with management specificities Blockchain and data mining Tracing and verifying reported landings by management authorities Adds transparency to fisheries data Facilitated enforcement of fisheries regulations Allows to direct inspection efforts to critical cases Difficult to implement and demanding on IT resources Algorithms may lead to wrong conclusions Sell at sea of catch by fishing companies to wholesalers factories, retailers Improved trading opportunities for producers (fishers) May incentivise the landing in countries with less strict management obligations Data mining Designation of RTC Improved speed in closure designation Requires near real-time transmission of catch information Data mining and AI Predictive control against IUUa, decision support for patrols Improved allocation of inspection efforts, higher incentive for fishers to comply with laws and management rules Requires near real-time transmission of catch information, which can still be flawed AI Catch recognition in on-board video footage Improved processing of catch, better data collection for management authorities Difficult to implement technically, may not be error or tamper-proof, high costs for companies operating fishing vessels Vessel identification of optical remote sensing images Improved knowledge on vessel location and activity May have a deterrent effect on illegal fishing vessels Spatial and temporal coverage may have significant gaps Error rate in vessel identification may be high Technology Task Strengths and opportunities Weaknesses and threats Blockchain Tracking of the supply chain of fish and seafood products by consumers and/or retailers Increase consumer trust and engagement in ecolabels and sustainability campaigns Not fraud proof as mislabelling/mal-codification still may occur Blockchain and smart contracts Trading of quotas/shares among fishing companies and between management authorities and fishing companies Transparent and self-organized trading of catch shares/quotas May not comply with management specificities Blockchain and data mining Tracing and verifying reported landings by management authorities Adds transparency to fisheries data Facilitated enforcement of fisheries regulations Allows to direct inspection efforts to critical cases Difficult to implement and demanding on IT resources Algorithms may lead to wrong conclusions Sell at sea of catch by fishing companies to wholesalers factories, retailers Improved trading opportunities for producers (fishers) May incentivise the landing in countries with less strict management obligations Data mining Designation of RTC Improved speed in closure designation Requires near real-time transmission of catch information Data mining and AI Predictive control against IUUa, decision support for patrols Improved allocation of inspection efforts, higher incentive for fishers to comply with laws and management rules Requires near real-time transmission of catch information, which can still be flawed AI Catch recognition in on-board video footage Improved processing of catch, better data collection for management authorities Difficult to implement technically, may not be error or tamper-proof, high costs for companies operating fishing vessels Vessel identification of optical remote sensing images Improved knowledge on vessel location and activity May have a deterrent effect on illegal fishing vessels Spatial and temporal coverage may have significant gaps Error rate in vessel identification may be high Note: This list is not intended to be comprehensive, but represents some potential applications together with associated strengths, weaknesses, opportunities, and threats. a Illegal, unreported, or unregulated. Problems and caveats Innovations are usually two-sided medals and hence blockchain, data mining, and AI pose challenges to all participants of fishing enterprises including fishermen, trader, consumers, management authorities, and scientist. Fishermen may not be willing to instigate further mechanisms of control, whereas traders and management authorities may fear the extra costs and effort of installing and maintaining new infrastructures. And finally, consumers may need to engage into the blockchain by downloading apps and spending time to get informed on their product. Furthermore, neither blockchains, smart contracts, nor data mining algorithms are free of fraud, error and uncertainty. It is therefore naïve to assume that these technologies will entirely prevent illegal or unreported fishing. One of the most crucial steps of the fisheries supply chain will remain the haul of the catch on board of the fishing vessel and the subsequent designation of labels. Both processes will still remain in the hands of the vessel crew. Outside the fishing vessel, it will be difficult to verify whether all catches are labelled correctly, whether discarding has occurred or whether catches are landed unreported or illegally. Labelling also becomes easier to manipulate when fish and seafood products are not sold as whole, but are processed into different product categories such as steaks, filets, loins and minced meat (Visser and Hanich, 2017). The technical infrastructure to maintain blockchains in near real-time is challenging. Receiving and sending large data volumes at sea is only possible via satellite communication in many parts of the ocean. Installing satellite communication devices may be not feasible in small-scale fisheries or fisheries with limited financial resources. Transmission prices could still be too high and the available bandwidth still too narrow to support the transmission demands of electronic logbooks and synchronised blockchains. Blockchains require intense and frequent communication between the nodes of the blockchain network. The synchronization of blockchains thus can be tardy (Christidis and Devetsikiotis, 2016) and may not work well in situations in which a large proportion of network nodes are faced with unstable network connections. Classical blockchains as used for Bitcoin are also faced with the challenge of scalability, i.e. a limited number of transactions that the network can process (Xu et al., 2016). Thus vast expansion of network participants may pose challenges on the blockchain network. Blockchains are also not fully tamper proof, as blockchain consensus could be reached by one party if it manages to hold the majority of the network nodes (Lin and Liao, 2017; Dey, 2018). While this may be an unlikely scenario in big blockchain networks, it may happen in smaller networks, when only a limited number of participants is involved. Predictive policing is commonly criticized for loss of privacy and civil rights, i.e. introducing ethnical bias into the identification of crime sites or individuals (Shapiro, 2017). Accordingly, fishermen may find their privacy interests breeched when being controlled by digital surveillance. Repeated protest against video monitoring at-sea and satellite-based VMSs in Europe indicate reluctance of fishermen to implement technologies which increase surveillance on their operations (Mangi et al., 2015). Thus data technologies should find their fastest and most widely accepted implementation in situations, in which fishermen are incentivised to do so, e.g. if they gain trade benefits or are relieved from paper work. Blockchains often require some sort of incentive for participation to maintain the network, which in cryptocurrencies is a mining reward. Obtaining this reward can be very energy consuming and costly, making an equitable participation of small and large stakeholders less likely. However, technical innovations of blockchain technologies are constantly produced and future applications may solve many of the aforementioned problems (Xu et al., 2016; Christidis and Devetsikiotis, 2016). For example, processing energy requirements can be reduced when blockchain systems are based on proof-of-stake algorithm instead of the proof-of-work algorithm (Christidis and Devetsikiotis, 2016; Lin and Liao, 2017). Proof-of-stake allocates the computation of new blocks in a deterministic way, e.g. based on the amount of currency held by a participant of the network. Contrary, proof-of-work is a competing scheme in which the potential creators of new blocks (i.e. miners) are competing by finding the fastest solution to a complex puzzle, which requires significant amounts of computational power and thus energy. Conclusions This article can only sketch some potential applications of blockchains, big-data analysis and AI in fisheries. Currently implemented examples and existing literature are too scarce to provide an in-depth review. Each technology will most likely evolve further, leading to unforeseen opportunities (and risks). Thus many of the described applications may never become realized, or their implementation may come with drawbacks which are not yet to be foreseen. But it is unlikely that the economics and management of fisheries will not be significantly affected by any of these technologies. Thus it is rather question of when and how enhanced data technologies will find entrance into the world’s fisheries. Blockchain, data mining, and AI will not stop IUU fishing, will not prevent overfishing and discarding. But they may help to make global streams of fish and seafood products with the associated flow of money becoming more visible and transparent. In fact, digital data technologies may work best in fisheries, which voluntarily intend to demonstrate their compliance to laws, management rules, and consumer demands or which are looking for a self-controlling mechanism to foster trust amongst competitors. Such systems may even evolve in areas, where governmental fisheries is currently weakly developed or totally absent, because fishermen may want to organize themselves to reduce conflicts and improve trade opportunities. Finally, in many situations, these technologies might allow governmental authorities to improve surveillance of industry compliance and consumers to place better informed decisions on which product they would like to purchase. Acknowledgements I thank Henrike and Phillip Rambo, who for the first time made me aware of the implications of crypto-currencies and blockchains. Two anonymous reviewers and the editors of ICESJMS provided valuable comments on the first version of the manuscript. References Benoit H. P. , Allard J. 2009 . Can the data from at-sea observer surveys be used to make general inferences about catch composition and discards? Canadian Journal of Fisheries and Aquatic Sciences , 66 : 2025 – 2039 . Google Scholar Crossref Search ADS Bo Y. , Ping S. , Huang G. 2012 . Development of traceability system of aquatic foods supply chain based on RFID and EPC internet of things . Transactions of the Chinese Society of Agricultural Engineering , 29 : 172 – 183 . Branch T. A. 2009 . How do individual transferable quotas affect marine ecosystems? Fish and Fisheries , 10 : 39 – 57 . Google Scholar Crossref Search ADS Cath C. , Wachter S. , Mittelstadt B. , Taddeo M. , Floridi L. 2018 . Artificial intelligence and the ‘good society’: the US, EU, and UK approach . Science and Engineering Ethics , 24 : 505 – 528 . Google Scholar PubMed Christidis K. , Devetsikiotis M. 2016 . Blockchains and smart contracts for the internet of things . IEEE Access , 4 : 2292 – 2303 . Google Scholar Crossref Search ADS Dey S. 2018 . A proof of work: securing majority-attack in blockchain using machine learning and algorithmic game theory . International Journal of Wireless and Microwave Technologies , 5 : 1 – 9 . Google Scholar Crossref Search ADS Dlodlo N. , Kalezhi J. 2015 . The internet of things in agriculture for sustainable rural development. In Proceedings of the International Conferecne on Emerging Trends in Networks and Computer Communications, Windhoek, Namibia, 17 May–20 May 2015. Institute of Electrical and Electronics Engineers (IEEE), Windhoek, Namibia. Dörner H. , Casey J. , Carvalho N. , Damalas D. , Graham N. , Guillen J. , Holmes S. J. et al. 2018 . Collection and dissemination of fisheries data in support of the EU Common Fisheries Policy . Ethics in Science and Environmental Politics , 18 : 15 – 25 . Google Scholar Crossref Search ADS Edgar G. J. , Bates A. E. , Bird T. J. , Jones A. H. , Kininmonth S. , Stuart-Smith R. D. , Webb T. J. 2016 . New approaches to marine conservation through the scaling up of ecological data . Annual Review of Marine Science , 8 : 435 – 461 . Google Scholar Crossref Search ADS PubMed FAO. 2016 . The State of World Fisheries and Aquaculture 2016. Contributing to Food Security and Nutritiona for All. Rome. 200 pp. Ghahramani Z. 2015 . Probabilistic machine learning and artificial intelligence . Nature , 521 : 452 – 459 . Google Scholar Crossref Search ADS PubMed Gulbrandsen L. H. 2009 . The emergence and effectiveness of the Marine Stewardship Council . Marine Policy , 33 : 654 – 660 . Google Scholar Crossref Search ADS Hall M. A. 1996 . On bycatches . Reviews in Fish Biology and Fisheries , 6 : 319 – 352 . Google Scholar Crossref Search ADS Haskell W. B. , Kar D. , Fang F. , Tambe M. 2014 . Robust detection of fisheries with COmPASS. In Proceedings of the Twenty-Sixth Annual Conference on Innovative Applications of Artfificial Intelligence, pp. 2978 – 2983 . Association for the Advancement of Artificial Intelligence, Québec City, Canada. Hastie T. , Tibshirani R. , Friedman J. ( 2009) The Elements of Statistical Learning, Springer Series in Statistics , Springer , New York . Haug S. , Castro R. P. , Kwon M. , Filler A. , Kowatsch T. , Schaub M. P. 2015 . Smartphone use and smartphone addiction among young people in Switzerland . Journal of Behavioral Addiction , 4 : 299 – 307 . Google Scholar Crossref Search ADS Helyar S. J. , Lloyd H. A. , de Bruyn M. , Leake J. , Bennett N. , Carvalho G. R. 2014 . Fish product mislabelling: failings of traceability in the production chain and implications for illegal, unreported and unregulated (IUU) fishing . PLoS One , 9 : e98691. Google Scholar Crossref Search ADS PubMed Herr H. , Fock H. O. , Siebert U. 2009 . Spatio-temporal associations between harbour porpoise Phocoena phocoena and specific fisheries in the German Bight . Biological Conservation , 142 : 2962 – 2972 . Google Scholar Crossref Search ADS Holmes S. J. , Campbell N. , Aires C. 2009 . Using VMS and fishery data in a real time closure scheme as a contribution to reducing cod mortality and discards. ICES Document CM 2009/M: 13. p. 27 . Hu J. , Li D. , Duan Q. , Han Y. , Chen G. , Si X. 2012 . Fish species classification by color, texture and multi-class support vector machine using computer vision . Computers and Electronics in Agriculture , 88 : 133 – 140 . Google Scholar Crossref Search ADS Jacquet J. L. , Pauly D. 2007 . The rise of seafood awareness campaigns in an era of collapsing fisheries . Marine Policy , 31 : 308 – 313 . Google Scholar Crossref Search ADS Kaiser M. J. , Clarke K. R. , Hinz H. , Austen M. C. V. , Somerfield P. J. , Karakassis I. 2006 . Global analysis of response and recovery of benthic biota to fishing . Marine Ecology Progress Series , 311 : 1 – 14 . Google Scholar Crossref Search ADS Kanjir U. , Greidanus H. , Ostir K. 2018 . Vessel detection and classification from spaceborn optical images: a literature survey . Remote Sensing of Environment , 207 : 1 – 26 . Google Scholar Crossref Search ADS PubMed Kantardzic M. ( 2011) Data Mining - Concepts, Models, Methods, and Algorithms , 2nd edn. Wiley , Hoboken, NJ . Kindt-Larsen L. , Kirkegaard E. , Dalskov J. 2011 . Fully documented fishery: a tool to support a catch quota management system . ICES Journal of Marine Science , 68 : 1606 – 1610 . Google Scholar Crossref Search ADS Kotsiantis S. B. 2007 . Supervised machine learning: a review of classification techniques . Informatica , 31 : 249 – 268 . Kourou K. , Exarchos T. P. , Exarchos K. P. , Karamouzis M. V. , Fotiadis D. I. 2015 . Machine learning applications in cancer prognosis and prediction . Computional and Structural Biotechnology Journal , 13 : 8 – 17 . Google Scholar Crossref Search ADS Kwon M. , Lee J.-Y. , Won W.-Y. , Park J.-W. , Min J.-A. , Hahn C. , Gu X. et al. 2013 . Development and validation of a smartphone addiction scale (SAS) . PLoS One , 8 : e56936. Google Scholar Crossref Search ADS PubMed Lewison R. , Crowder L. , Read A. , Freeman S. 2004 . Understanding impacts of fisheries bycatch on marine megafauna . Trends in Ecology & Evolution , 19 : 598 – 604 . Google Scholar Crossref Search ADS Li S. , Xu L. D. , Zhao S. 2015 . The internet of things: a survey . Information Systems Frontiers , 17 : 243 – 259 . Google Scholar Crossref Search ADS Lin I.-C. , Liao T.-C. 2017 . A survey of blockchain security issues and challanges . International Journal of Network Security , 19 : 653 – 659 . Mangi S. C. , Dolder P. J. , Catchpole T. L. , Rodmell D. , de Rozarieux N. 2015 . Approaches to fully documented fisheries: practical issues and stakeholder perceptions . Fish and Fisheries , 16 : 426 – 452 . Google Scholar Crossref Search ADS Mora C. , Myers R. A. , Coll M. , Libralato S. , Pitcher T. J. , Sumaila R. U. , Zeller D. et al. 2009 . Management effectiveness of the world’s marine fisheries . PLoS Biology , 7 : e1000131. Google Scholar Crossref Search ADS PubMed Mosler H.-J. 1993 . Self-dissemination of environmentally responsible behavior: the influence of trust in a common dilemma game . Journal of Environmental Psyhology , 13 : 111 – 123 . Google Scholar Crossref Search ADS Needle C. L. , Catarino R. 2011 . Evaluating the effect of real-time closures on cod targeting . ICES Journal of Marine Science , 68 : 1647 – 1655 . Google Scholar Crossref Search ADS Pauly D. , Zeller D. ( 2003) The global fisheries crisis as a rationale for improving the FAO’s database of fisheries statistics. In From Mexico to Brazil: Central Atlanitc Fisheries Catch Trends and Ecosystem Models , pp. 1 – 9 . Fisheries Centre Research Reports. Ed. by Zeller D. , Booth S. , Mohammed E. , Pauly D. . University of British Columbia, Vancouver, Canada. Pearsall B. ( 2013) Predictive policing: the future of law enforcement? National Institute of Justice Journal , 266 : 16 – 19 . Pfreundt U. 2018 . How to harness blockchain technology for marine conservation . PeerJ Preprints , 6 : e26496v2 . Rajamaran V. 2016 . Big data analytics . Resonance , 21 : 695 – 716 . Google Scholar Crossref Search ADS Russell S. J. , Norvig P. ( 2010) Artificial Intelligence: A Modern Approach . Pearson , New York , 37 pages. Russom P. 2011 . Big data analytics . TDWI Best PRactive Report , 37 . Ruths D. , Pfeffer J. 2014 . Scoial media for large studies of behavior . Science , 346 : 1063 – 1064 . Google Scholar Crossref Search ADS PubMed Samaha M. , Hawi N. S. 2016 . Relationships among smartphone addiction, stress, academic performance, and satisfaction with life . Computers in Human Behavior , 57 : 321 – 325 . Google Scholar Crossref Search ADS Schröder U. 2008 . Challanges in the traceability of seafood . Journal für den Verbraucherschutz und Lebensmittelsicherheit , 3 : 45 – 48 . Google Scholar Crossref Search ADS Shapiro A. 2017 . Reform predictive policing . Nature , 541 : 458 – 460 . Google Scholar Crossref Search ADS PubMed Storbeck F. , Daan B. 2001 . Fish species recognition using computer vision and neural network . Fisheries Research , 51 : 11 – 15 . Google Scholar Crossref Search ADS Stransky C. , Berkenhagen J. , Berth U. , Ebeling M. , Jiménez-Krause J. D. , Panten K. , Schultz N. 2008 . National fisheries data collection programme: activities and outlook . Informationen aus der Fischereiforschung , 55 : 5 – 14 . Sumaila U. R. , Marsden A. D. , Watson R. , Pauly D. 2007 . A global ex-vessel fish price database: construction and applications . Journal of Bioeconomics , 9 : 39 – 51 . Google Scholar Crossref Search ADS Swan M. ( 2015) Blockchain: Blueprint for a New Economy . O’Reilly Media , Sebastopol, CA . Tapscott D. , Tapscott A. ( 2016) Blockchain Revolution: How the Technology behind Bitcoin Is Changing Money, Business, and the World . Penguin , New York . Tayal D. K. , Jain A. , Arora S. , Agarwal S. , Gupta T. , Tyagi N. 2014 . Crime detection and criminal identification in India using data mining techniques . AI & Society , 30 : 117 – 127 . Google Scholar Crossref Search ADS Ulleweit J. , Stransky C. , Panten K. 2010 . Discards and discarding practices in German fisheries in the north Sea and Northeast Atlantic during 2002–2008 . Journal of Applied Ichthyology , 26 : 54 – 66 . Google Scholar Crossref Search ADS Urmson C. , Anhalt J. , Bagnell D. , Baker C. , Bittner R. , Clark M. N. , Dolan J. et al. 2008 . Autonomous driving in urban environments: boss and the urban challenge . Journal of Field Robotics , 25 : 425 – 466 . Google Scholar Crossref Search ADS van Helmond A. T. M. , Chen C. , Poos J. J. 2014 . How effective is electronic monitoring in mixed bottom-trawl fisheries? ICES Journal of Marine Science , 72 : 1192 – 1200 . Google Scholar Crossref Search ADS Visser C. , Hanich Q. A. 2017 . How blockchain is strengthening tuna tracebility to combat illegeal fishing . The Conversation , 4 , 4 pages. Walker S. J. 2014 . Big Data: a revolution that will transform how we live, work, and think . International Journal of Advertising , 33 : 181 – 183 . Google Scholar Crossref Search ADS Xu X. , Pautasso C. , Zhu L. , Gramoli V. , Ponomarev A. , Chen S. 2016 . The blockchain as a software connector. In Proceedings of the 13th Working IEEE/IFIP Conference on Software Architecture (WICSA) Venice, Italy, 5–8 April 2016. Institute of Electrical and Electronics Engineers (IEEE), Venice, Italy. Zakir J. , Seymour T. , Berg K. 2015 . Big data analytics . Issues in Information Systems , 16 : 81 – 90 . © International Council for the Exploration of the Sea 2019. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - How emerging data technologies can increase trust and transparency in fisheries
JF - ICES Journal of Marine Science
DO - 10.1093/icesjms/fsz036
DA - 2019-03-14
UR - https://www.deepdyve.com/lp/oxford-university-press/how-emerging-data-technologies-can-increase-trust-and-transparency-in-xV11yFIRpL
SP - 1
VL - Advance Article
IS - 
DP - DeepDyve
ER -