Boyd Tarlinton, Yiru Chen, Neil White, Stuart Irvine Brown, Geoff Dickinson, Dario Stefanelli and Daryl Joyce from DPI and DPIRD.
Hold tight - we’re checking permissions before loading more content
The horticultural sector at large is embracing data and digital transformation.
In this regard, the avocado industry is increasingly recognising data-driven decision making as valuable.
Industry operatives who routinely capture, manage, and analyse their data derive profitable insights.
Hort Innovation project AV21005 – ‘Growing robust avocados’ involves field trials at eight sites across Australia over three years with supplementary field and laboratory trials.
These trials all produce datasets that can be used to explore genetic, environmental, and management factor influences on fruit robustness.
Lessons from our implementation of a project database can inform avocado producers, supply chain partners, and other industry stakeholders including research, development and extension workers interested in and willing to embrace the data age.
The potential benefits of storage, accessibility, and arrangement of data for scientific and technical investigation and profit-focused decision making are rapidly and broadly being realised in horticulture.
Embracing the data age underpins state-of-the-art technologies, including generative AI and farm automation.
Yet much data is currently trapped in spreadsheets, digital documents, and paper records.
Opportunity exists to turn legacy datasets to competitive advantage by unifying information, removing duplication, and facilitating future record keeping to streamline actionable analyses.
AV21005 is one of the avocado industry’s most data-intensive paddock to plate projects to date, necessitating an open-minded approach to data.
While storing data in spreadsheets has historically been effective, discrepancies often appear between data collected at different sites and over subsequent years of individual projects.
Moreover, difficulties emerge between projects when re-analysing data, project members move on, or records are lost.
We approached AV21005 with a view to not just facilitate data collection and analysis for this project, but to build a database with a legacy for the long-term, including future industry research, development, and extension activities.
Avocado industry members likely already interact with databases in their professional capacity.
Examples in the horticulture industry include the TradeAtlas Database from which Hort Innovation’s Australian Horticulture Statistics Handbook is derived (TradeAtlas, 2025), the Australian Pesticides and Veterinary Medicines Authority’s (APVMA) Public Chemical Registration Information System database (APVMA, 2025), and the sustainability-related Regulations Database provided by the Australian Farm Institute (AFI, 2025).
These resources illustrate the value embodied in a database, including by allowing users to quickly perform complex queries against a central source of truth.
AV21005 aims to bring this ease of access, utility, and certainty to data captured on-farm and downstream through the supply chain for this project and those in the future.
In AV21005 we adopted the Katmandoo database system to manage our data.
Katmandoo was developed by the Queensland Government Department of Primary Industries and NSW Department of Primary Industries and Regional Development (DPI, 2025) using the free and open-source relational database management system MySQL.
We deployed it on a server at the DPI Gatton Smart Farm for full oversight over grower data.
Foremost among considerations were customisability, reliability, and initial accessibility for research users with a range of backgrounds.
It was populated with data from three separate Australian avocado growing regions, tropical far north Queensland, sub-tropical central Queensland, and Mediterranean south-west Western Australia.
Although seemingly complex, the AV21005 project database allows seamless integration and query of preharvest agronomy data, postharvest outcomes, trial metadata, and geospatial data.
This allows users to access a wide variety of data, test different theories of growing robust avocados, and answer questions posed in the context of growing robust avocados.
We can also act directly on data we collect; graphing mineral nutrient contents in soil solutions and leaf or fruit tissues allows visual interpretation towards informed decision making.
Building additional tools around such a database allows data to be viewed and understood by a broad range of users.
A dashboard can provide quick visual summaries of data collected to provide ready insights into supply chain outcomes.
Dashboards can be produced with tools like Tableau or Microsoft’s Power BI or built from the ground up.
Some growers may prefer a commercial data management solution that integrates with their existing business management software.
An organisation must also consider whether to host a database locally or make use of a cloud platform.
Investing in a local solution may require a greater immediate outlay in terms of infrastructure and development costs.
Cloud-based services may require less investment initially but can lock an organisation into ongoing and sometimes complex management fees.
Regardless of the platform, regular backups must be a primary imperative.
The FAIR standards are a commonly applied set of principles for database management, necessitating data to be findable, accessible, interoperable, and reusable (Wilkinson et al., 2016).
There is an ideal to ensure that data is discoverable such that requests for access can be made where appropriate regardless of whether the database itself is private.
While this aim applies primarily to research databases, lessons can be applied to benefit industries.
Two agencies collecting the same information can represent senseless duplication of effort versus efficient and effective data management.
Ultimately, databases unlock the power of data for productive and profitable decision making to both identify immediate interventions and reveal trends for remedial actions, such as around site-specific quantities and timings of fertiliser applications.
Starting with a deep understanding of data collected, and posing meaningful questions of databases can inform fit-for-purpose solutions.
Experience from AV21005 underscores the value and importance to industry of thoughtful database design with a view to identify and extract maximum information from genetic, environmental, and management practice data as the Australian avocado industry pioneers the growing of robust avocados to meet sustainable export market needs.
Database Terms to Know
- Big Data: Exceptionally large datasets, often requiring specialised databases. Big Data is typically characterised by volume (the amount of data), velocity (the speed at which data must be processed), and variety (the complexity of the data). Examples of Big Data may include tracking avocado consumer behaviour and shopping habits to deliver hyper-personalised retail avocado product recommendations tailored to individual customers upon request, or in monitoring historical customer avocado product purchasing patterns and analysing them against grower production activity to inform agile and balanced supply/demand pricing.
- Cloud: The term ‘Cloud’ describes applications and databases which are managed by an external provider and accessed over the internet. The Cloud typically overcomes many of the complexities involved in managing a service locally. Prominent Cloud services include Amazon Web Services (AWS) and Microsoft’s Azure platform, both of which offer many application and database services.
- Dashboard: An application interface used to visualise, display and derive insights from the data stored in a database. These can be simple, prebuilt applications, or more complex custom applications designed to suit a specific data display purpose. You might want to visualise mineral concentrations in your soils or watch for a spike in defective fruit.
- IoT: The ‘Internet of Things’ describes a paradigm where a multitude of physical internet-connected devices provide access to real-time data. Capturing and interpreting this data can be challenging. In horticulture, examples include weather stations, data from packing lines, and real-time supply chain monitoring.
- Table: Rows typically correspond to some sort of observation or entity, while columns represent a descriptor. For example, a table used to store soil nutrient analyses might have a row for each sample, and a column for each nutrient measured.
- Query: A query is a piece of computer language code that asks a question of a database. A well-written query can distil a complex dataset down to an interpretable summary, for example finding the characteristics of trees with the highest yield in an orchard.