Efficient farm management, good scientific research, and effective selection are possible only if the data are easily accessible, accurate, and gathered routinely. The upkeep of paper records is not only cumbersome and difficult to maintain but is also hard to replicate and requires a lot of physical space. The lack of proper digitized farm records has hindered farm development as well as the creation of future-oriented breeding policies. Also, the absence of a centralized cloud-based database and the unavailability of duplicate records leads to the possibility of data loss due to factors like pests, theft, loss, fire, floods, etc. which have happened in some instances in the past. Progress in the farm operation cannot be determined from year to year without keeping an inventory and deriving rapid inferences from paper records is impractical but a digital inventory is flexible, efficient, rapid, and accurate32. A web-based DSS catering to most aspects of sheep farm data management would help in the creation of a system for the safe storage, analysis, retrieval, and effective use of data.
It is almost impossible to draw real-time, on-spot inferences from manual data. This especially hinders breeding decisions that require procedures that are computationally expensive and require vast amounts of data. Having to manually enter massive amounts of data into the system before analysis consumes a considerable amount of a researcher’s time and energy33. As a result, research is not only delayed but the quality of the actual research is compromised. Therefore, selection on farms in India is still intuition based and the genetic progress is very low. The benefits in terms of saving time by efficient data handling have also been stressed by Lawson et al.34.
E-linking of the University farm with other sheep farms would be necessary for ONBS which would generally increase the genetic merit of animals across farms35. Web-based farm management information systems facilitate internet-based collaborative research by connecting geographically dispersed farms with experts or customizing end-user data for analysis or presentation36.
As also suggested by Fountas et al.37 agriculture has entered a new era in which the purpose of a DSS is to meet the increased demands, reduce production costs, and increase the overall productivity of the farm38. Government farm managers and officers of the Sheep Husbandry Department had felt the need for this tool and their complete agreement upon sensitization in its utility enforced the hypothesis of this research. The key to success is access to timely information and elaborate decision-making. Decision-making is an important aspect of farm management as well as e-governance. Similar tools for dairy cattle have been developed8. The most popular one among them is INAPH, launched countrywide by the Government of India. Commercially available superior quality tools for sheep farm management were found to be scanty and inadequate. Additionally, they are complex and difficult to use39 and are not applicable to farm practices and present systems of data collection on Indian farms. Adoption of such systems would force such farms to alter some of their processes to fully utilize the product thus adding to the problem rather than subtracting from it40. This would also reduce adoption rates substantially. SSB solves this issue by being custom tailored and specifically designed for challenges faced by farmers in India.
Additionally, the available data management software capable of estimating genetic parameters for ranking animals according to their genetic merit is not easily available and those that are, do not assist in making breeding-related decisions. This necessitated the need for the development of a custom-tailored and multi-dimensional DSS for catering to the needs of the sheep farms. Economic, as well as scientific progress, could be ensured by adopting the present tool because it is simple to use, easy to understand, and cheap. Also, the present tool was developed with a user-friendly interface for sheep farms. Automated data processing would also facilitate decision-making in real-time, futuristic, and effective policy planning for breeding by administrators, and scientific research on the farm and allow the integration of expert knowledge and farming preferences for the benefit of both. Managemental strategies could be improved using this tool through proper interpretations of the trend/status of the farms. Smart Sheep Breeder is a step toward precision animal farming which has also been reported to increase efficiency, reduced labor and improve overall animal well-being41.
With the University as a stakeholder, farmers would generally be more comfortable with sharing data than with large companies that may repurpose the data for corporate interests. This is also a primary concern of India’s digital agriculture mission.
The success of animal farming in developed countries may mainly be attributed to the incorporation of specific DSS and breeding tools. Sheep Genetics Australia, the National Genetic Information and evaluation services for the meat and wool sector, delivered as LAMBPLAN, MERINOSELECT hosts a database of about 6.5 million animals and more than 1000 flocks in Australia and overseas.
Roles assigned to specific users ensure that relevant data is served to the users as per their role and ensure privacy for farmers. For e-governance purposes i.e., for the remote monitoring of all farms under his jurisdiction, an administrator needs to have access to data across all farms to get a general overview of all farms. He/she could thus remotely monitor all farms under the jurisdiction, visualize the farming trends in real-time and be able to take effective data-backed decisions. All this must be achieved without compromising the accuracy of the records from other farms thereby maintaining the privacy and hassle-free use of the web-based tool.
The most important function of a DSS is to handle and benefit from enormous data volumes and convert raw data into useful information39. These would give the farm manager an understanding of the farm situation and the administrator to take timely critical decisions related to all aspects of farm management. Farm reports serve as an aid to managerial control during production. A producer can keep track of events such as whether activities are going according to plan, the total strength of the farm in terms of animals according to their sex and age, when animals were vaccinated, dipped, given any medicine, or castrated, which breed on the farm is performing better and which breed is not and production from animals both in terms of quality and quantity may be effectively monitored. Inferences drawn from disposal reports may give an early indication of any potential problems at the farm e.g., epidemic, managemental errors, etc. An analysis of reports related to breeding may help the breeder to find out expected dates of parturition, help in removing sterile/infertile animals, and make use of superior quality rams for breeding. History sheets of individual animals provide all information about that animal which helps the breeder to draw inferences and make decisions regarding breeding/culling/sale of the animal.
A breeder may also trace the pedigree of animals and serve as a tool for the selection of breeding animals and refrain from mating animals that are closely related to each other. A sound breeding program can only be developed by understanding which pedigree to use based on the information collected about each ancestor.
The tool was validated by both alpha and beta testing. The similarity of the results obtained by the DSS with book values and standard tools proves its accuracy. The retrospective data was used so that the tool could be validated quickly. Since the tool was developed to eventually be used on real-world farms, actual farm data, both real-time as well as retrospective was used to validate the tool. The results produced by the system could thus be compared with true data and they were found to be accurate. E.g., The lambing rate, animal parities, disposal rates, and farming trends were found to be 100% accurate. Results that could not be checked against farm records e.g., breeding values estimations were compared with standard methodologies or standard software like Wombat. The alpha testing and the feedback obtained through beta testing results validated the initial hypothesis.
The use of AI and various biometrical techniques for the ranking of animals through estimated breeding values and selection indices shall enable farm managers to retain only animals of high genetic merit on the farm and cull animals of low genetic merit and thus ensuring that genetic merit is maintained at the farm (Fig. 3).
The estimation of breeding values using the mixed model method for BLUP provides a powerful and flexible tool. Artificial Intelligence is also becoming a significant change in the agricultural sector. Deep learning models have also been used for sheep behavior recognition42. Lopes et al.43 also showed that the ANNs performed better than other ML techniques for the prediction of genetic merit. AI is also finding applications in wide-ranging areas of research44,45 including systems for providing decision support46.
A higher correlation of 0.89 than that obtained by the present study, using ANN was reported for breeding values prediction by Bangar et al.47 in the Harnali sheep breed. Ghotbaldini et al.48 for the prediction of breeding values in Kermani sheep also used two ANN models and reported correlation coefficients of 0.703 and 0.864. Our results are also consistent with the reports of other scientists in this area49,50. Also, a low learning rate was found to give better results than a higher one by Brownlee51 because high learning rates may cause unstable training and the neural network may never actually converge. IoT-based tools are rapidly becoming a norm in the world52. This is true also for the animal sciences. Cheng et al.42 also suggest the use of sensors, and machine learning, monitoring for monitoring animals and farms through automation which is necessary as farming is becoming more and more intensive. The AI model used would enable the farmers to know the genetic superiority of their animals at their fingertips and therefore make the right breeding decisions.
There are no major ethical issues involved with the adoption of this tool as no animal interventions are performed by this. Additionally, the IoT compatibility of the system is to ensure that animal handling is minimized and that the drudgery of data entry is eliminated. Most farms in J&K today are based on manual weighing systems which are labor-intensive and involve a tedious ledger-based record-keeping system. The development of a technology-driven system for farm management would promote both farm and animal welfare (as the animals do not need to be lifted by a rope to be hung from the spring balance for weighing anymore). The ultimate goal of developing this system is to minimize handling animals which would reduce animal distress and as farming becomes resourceful, more animals could be raised by a single farmer without additional labor.
A major threat to validity is the use of spurious data. This has been prevented for the weight records by the use of IoT-based data entry operations. In addition, the tool uses various checks like limiting entries to format types or setting upper and lower limits for value entries to ensure that outlier data is not entered into the system. This could be further improved by using AI for data validation as well. The threat of low adoption has been mitigated to a large extent by ensuring the system is simple to use and easily accessible. Also, the steady penetration of the internet and mobile phones to remote corners of the Country would help in the popularization of the tool. This is also in line with the Digital India mission of the Govt. of India.
BLUP has become the worldwide standard for the prediction of breeding values of farm animals. This is because, this method minimizes the variances of errors, the correlation between predictors and predictions are higher, the probability of selecting the better of any pair of candidates is maximized and the expected mean of the breeding values of selected individuals is maximized. The DSS also generates inbreeding coefficients like a tool reported by53, which may help the farm manager to monitor inbreeding on the farm and make necessary decisions regarding the import of new animals into the farm or culling of inbred animals whenever required.
The generated report on selection index values of all live animals on the farm for the DSS shall serve as a breeding tool to make the selection on more than one trait at once. Since in practice, the index value is a weighted BLUP-EBV of selection candidates, therefore, the complexity of b = P–1Gv was avoided. The weighted sum of EBV and (economic) breeding goal weights for the construction of an index are also used in the SelAction tool, LAMBPLAN, and WOOLPLAN as reported by27. A single selection index may not serve the interest of all farms because each producer may have different feed costs and sales markets so an index specifically for that flock would be better than a one-size-fits-all index.