Addressing Data Bottlenecks in the Dairy Farm Industry.

A survey to explore the challenges and opportunities for dairy farm data management and governance was completed by 73 farmers and 96 non-farmers. Although 91% of them find data sharing beneficial, 69% are unfamiliar with data collection protocols and standards, and 66% of farmers feel powerless over their data chain of custody. Although 58% of farmers share data, only 19% of them recall having signed a data share agreement. Fifty-two percent of respondents agree that data collected on farm belongs only to the farmer, with 25% of farmers believing intellectual property products are being developed with their data, and 90% of all said companies should pay farmers when making money from their data. Farmers and non-farmers are somewhat concerned about data ownership, security, and confidentiality, but non-farmers were more concerned about data collection standards and lack of integration. Sixty-two percent of farmers integrate data from different sources. Farmers' most used technologies are milk composition (67%) and early disease detection (56%); most desired technologies are body condition score (56%) and automatic milking systems (46%); most abandoned technologies are temperature and activity sensors (14%) and automatic sorting gates (13%). A better understanding of these issues is paramount for the industry's long-term sustainability.

Data Governance in the Dairy Industry.

Data governance is a growing concern in the dairy farm industry because of the lack of legal regulation. In this commentary paper, we discuss the status quo of the available legislation and codes, as well as some possible solutions. To our knowledge, there are currently four codes of practice that address agriculture data worldwide, and their objectives are similar: (1) raise awareness of diverse data challenges such as data sharing and data privacy, (2) provide data security, and (3) illustrate the importance of the transparency of terms and conditions of data sharing contracts. However, all these codes are voluntary, which limits their adoption. We propose a Farmers Bill of Rights for the dairy data ecosystem to address some key components around data ownership and transparency in data sharing. Our hope is to start the discussion to create a balanced environment to promote equity within the data economy, encourage proper data stewardship, and to foster trust and harmony between the industry companies and the farmers when it comes to sharing data.

Symposium review: Dairy Brain-Informing decisions on dairy farms using data analytics.

Management decisions can be informed by near-real-time data streams to improve the economics of the farm and to positively benefit the overall health of a dairy herd or the larger environment. Decision support tools can use data management services and analytics to exploit data streams from farm and other economic, health, and agricultural sources. We will describe a decision support tool that couples data analytics tools to underlying cow, herd, and economic data with an application programming interface. This interface allows the user to interact with a collection of dairy applications without fully exposing the intricacies of the underlying system model and understand the effects of different decisions on outputs of interest. The collection of these applications will form the basis of the Dairy Brain decision support system, which will provide management suggestions to farmers at a single animal or farm level. Dairy operations data will be gathered, cleaned, organized, and disseminated through an agricultural data hub, exploiting newly developed ontologies for integration of multiple data sources. Models of feed efficiency, culling, or other dairy operations (such as large capital expenditures, outsourcing opportunities, and interactions with regulators) form the basis of analytical approaches, operationalized via tools that help secure information and control uncertainties. The applications will be independently generated to provide flexibility, and use tools and modeling approaches from the data science, simulation, machine learning, and optimization disciplines to provide specific recommendations to decision makers. The Dairy Brain is a decision support system that couples data analytics tools with a suite of applications that integrate cow, herd, and economic data to inform management, operational, and animal health improving practices. Research challenges that remain include dealing with increased variability as predictions go from herd or pen level down to individual cow level and choosing the appropriate tool or technique to deal with a specific problem.

Minimizing opportunity costs to aquatic connectivity restoration while controlling an invasive species.

Controlling invasive species is critical for conservation but can have unintended consequences for native species and divert resources away from other efforts. This dilemma occurs on a grand scale in the North American Great Lakes, where dams and culverts block tributary access to habitat of desirable fish species and are a lynchpin of long-standing efforts to limit ecological damage inflicted by the invasive, parasitic sea lamprey (Petromyzon marinus). Habitat restoration and sea-lamprey control create conflicting goals for managing aging infrastructure. We used optimization to minimize opportunity costs of habitat gains for 37 desirable migratory fishes that arose from restricting sea lamprey access (0-25% increase) when selecting barriers for removal under a limited budget (US$1-105 million). Imposing limits on sea lamprey habitat reduced gains in tributary access for desirable species by 15-50% relative to an unconstrained scenario. Additional investment to offset the effect of limiting sea-lamprey access resulted in high opportunity costs for 30 of 37 species (e.g., an additional US$20-80 million for lake sturgeon [Acipenser fulvescens]) and often required ≥5% increase in sea-lamprey access to identify barrier-removal solutions adhering to the budget and limiting access. Narrowly distributed species exhibited the highest opportunity costs but benefited more at less cost when small increases in sea-lamprey access were allowed. Our results illustrate the value of optimization in limiting opportunity costs when balancing invasion control against restoration benefits for diverse desirable species. Such trade-off analyses are essential to the restoration of connectivity within fragmented rivers without unleashing invaders.

Quantifying invasion resistance: the use of recruitment functions to control for propagule pressure.

Invasive species distributions tend to be biased towards some habitats compared to others due to the combined effects of habitat-specific resistance to invasion and non-uniform propagule pressure. These two factors may also interact, with habitat resistance varying as a function of propagule supply rate. Recruitment experiments, in which the number of individuals recruiting into a population is measured under different propagule supply rates, can help us understand these interactions and quantify habitat resistance to invasion while controlling for variation in propagule supply rate. Here, we constructed recruitment functions for the invasive herb Hieracium lepidulum by sowing seeds at five different densities into six different habitat types in New Zealand's Southern Alps repeated over two successive years, and monitored seedling recruitment and survival over a four year period. We fitted recruitment functions that allowed us to estimate the total number of safe sites available for plants to occupy, which we used as a measure of invasion resistance, and tested several hypotheses concerning how invasion resistance differed among habitats and over time. We found significant differences in levels of H. lepidulum recruitment among habitats, which did not match the species' current distribution in the landscape. Local biotic and abiotic characteristics helped explain some of the between-habitat variation, with vascular plant species richness, vascular plant cover, and light availability, all positively correlated with the number of safe sites for recruitment. Resistance also varied over time however, with cohorts sown in successive years showing different levels of recruitment in some habitats but not others. These results show that recruitment functions can be used to quantify habitat resistance to invasion and to identify potential mechanisms of invasion resistance.