Endemic to Aotearoa, the hoiho is one of the world’s rarest penguin species. This nationally-endangered penguin sparked intrigue and concern for Dr. Stefan Saverimuttu after ‘mass die-off’ events coincided with a mainland population decrease of 65% in the last 20 years.

Hoiho, more commonly known as the yellow-eyed penguin, and kororā (little penguin) face a similar range of threats, from predators; human disturbance; habitat and marine degradation; and disease. Current technologies are lacking in their capability to leverage current data and look for broad trends – a problem described by Stefan as a ‘global issue’. In collaboration with Murdoch University, Massey University, and Auckland Zoo, Stefan is creating a computer program that can easily leverage modern databases to describe broad trends and help conservationists understand the current and most imminent problems facing our native wildlife – starting with our endemic penguin species.

Every native species that dies in New Zealand gets a post-mortem (where possible) with the results uploaded into a database managed by Wildbase Pathology at Massey University. As Stefan points out, there is a tremendous amount of effort going into collecting, maintaining and collating this data, which is incredibly useful for looking up individual cases. Where Stefan sees further opportunity is in using this valuable data to create a tool that can further assist national conservation efforts.

Whilst the research project started out as a post-mortem review of penguin species to broadly gage morbidity trends and understand prominent diseases in hoiho like diphtheritic stomatitis, COVID-19 restrictions created extra challenges when it came to visiting laboratories. This led Stefan to explore a fresh angle for this research project: the ability to utilise pre-existing databases to build a computer program that can read the data and illustrate broad die-off trends. The importance of being able to access these trends quickly, especially in a country where many of our native species are in decline, is unparalleled.

The importance of being able to access these trends quickly, especially in a country where many of our native species are in decline, is unparalleled.

Dr. Stefan Saverimuttu

The initial stage of the research project involved 1,394 individual penguin post-mortem reviews – every penguin post-mortem stored on the Massey Pathology database, with over 800 of these being hoiho reviews alone. This information was then transcribed into a separate spreadsheet with standardised terminologies, essentially making it analysable for Stefan to be able to publish his findings in a research paper, broadly describing the current trends leading to penguin deaths in New Zealand. The next stage of his research project involves the creation of a program to be used in New Zealand, that will essentially be able to do this review automatically.

"I had to learn some statistics, and how to use a program called ‘R’ widely used in the biological sciences, and became to understand a concept called ‘text mining’, where a computer will read and pull a report for you," Stefan explained. 

Text mining is a really helpful tool for understanding broad trends. For example, with where the computer program is currently at, the main key words that show up in relation to kororā (little penguin) are ‘emaciation’, ‘oiling’, ‘bite’, ‘mustelid’, ‘puncture’, so the computer can help create a picture of what it happening to kororā, and analysis can be undertaken. This could give conservationists the tools to determine if more mustelid control needs to be done, or there is a trend of starving penguins starving – indicating potentially a lack of food source, or competition for food.

The concept is to ensure that any native species, be it kākāpō, takahē, or a New Zealand sea lion, is able to have all post-mortem data captured on a program where a computer can read the information, and output broad trends, without a researcher specifically needing to spend weeks reading reports to write a manual review. This kind of information is imperative to understanding the leading cause of death in any native species, and then being able to give conservationists the tools to mitigate these risks, and change declining population trends.

The final stage of this research project involves comparing the broad trends identified in the manual post-mortem review, with the broad trends identified by Stefan’s text-mining program, to validate the program’s results. Following this, manual post-mortem reviews of other species will be run through the program, with results compared, validating the effectiveness of the program. Stefan hopes that by publishing this code, and being able to pass this on to New Zealand conservationists, this is something that can contribute to conservation in Aotearoa for a long time to come.

Dr. Stefan Saverimuttu first studied veterinary medicine at the University of Sydney, before gaining five years’ experience in mixed practice clinics from New South Wales to the Northern Territory. Stefan is completing a Doctorate of Veterinary Medical Science at Murdoch University, whilst undertaking a residency based at Auckland Zoo.