Modelling the natural world

DISTINGUISHED PROFESSOR MARTI ANDERSON - New Zealand Institute for Advanced Study, College of Sciences

Nature is a complex collection of ecosystems, each containing many species. To understand how ecosystems are changing, statistical methods — abstract, mathematical representations of ecological systems — are often employed.

Using data gathered from the field, these methods can be used to characterise an ecosystem. This allows predictions to be made about how ecosystems might respond to proposed human activity, showing potential environmental impacts.

Traditional approaches have tended to concentrate on one species at a time, because modelling more than this is extremely challenging mathematically. Although these can give useful information about a single species, they are unable to show how large numbers of species act together, so they cannot be used to model the behaviour of a real ecosystem.

To help with this, Distinguished Professor Marti Anderson of the New Zealand Institute for Advanced Study has spent her career finding ways to improve statistical approaches for modelling ecosystems. ‘I realised that the statistical methods that were needed didn’t exist, so I started to look at developing my understanding of newer types of methods that were more computer-intensive. I was able to develop new methods that would work not just for my data but also for anybody’s data who was working in ecology or the environment.

‘My work is quite broad. I publish formal mathematical statistical methods, but these can be applied to any ecological system that has multiple variables, from trees in a forest to microbes in the soil. There are lots of really interesting contexts where these methods are being used and new insights are being brought to life because of these new tools.’

As well as her work in developing new mathematical methods, Distinguished Professor Anderson is also an active ecologist. Currently, she is working on a joint project with the Museum of New Zealand Te Papa Tongarewa, funded by a Marsden grant, to learn more about how the biodiversity of fishes changes as the ocean becomes deeper. Biodiversity is known to be affected by environmental extremes; for example, near the poles species become fewer than near the equator, where conditions are more favourable. A similar situation is seen with altitude — from sea level to a high mountain peak, the environment gradually becomes more variable and more extreme, affecting biodiversity.

I am really interested in these big-scale biodiversity questions: What are the different forms that life takes? What are the functions that they perform? How are they related in evolutionary time? What sort of threats do they have?


The current project is looking at how biodiversity is affected by extremes that occur in the depths of the ocean where, for example, light is scarce. As part of this, PhD student Lizzy Myers is developing a database containing measurements of fishes, from shallow waters to depths of up to 1200 metres, to see how they are changing with depth. Concurrently, postdoctoral researcher Dr David Eme is developing a new phylogenetic tree for New Zealand marine fishes to enrich our understanding of the emergence of fish traits through time.

‘I am really interested in these big-scale biodiversity questions,’ says Distinguished Professor Anderson. ‘It’s moving from just counting how many species we’ve got to asking questions like: What are the different forms that life takes? What are the functions that they perform? How are they related in evolutionary time? What sort of threats do they have? Those kinds of questions are going to be underpinned by a lot of fundamental science. I am thinking about how we can include functional information and shared evolutionary ancestries in our modelling.’

As well as her academic work, Distinguished Professor Anderson is also the director of software company PRIMER-e, which produces ecological modelling software widely used by experts in a variety of fields, from ecologists to government agencies such as NIWA, Landcare Research, the Environmental Protection Authority, and the National Oceanic and Atmospheric Administration in the United States, as well as private environmental consultants. The company also runs specialist workshops on using the software.

‘I just love learning about nature and I love interacting with people to help them gain new insights,’ she says. ‘I get emails from people all over the world every day asking for advice about analysing their data, and they are often working in really interesting systems; for example, analysing pumice stones that were shooting out of volcanoes, floating in the ocean and then washing up on a beach. They collect the pumice, see what critters have colonised it, and try to work out its origin and its travel though time. In this field, you find this beautiful conversation between the ecologist who is asking the right question and the statistician who can help them answer it; then there is the statistician who, in turn, develops a new method that allows the ecologist to answer questions they haven’t been able to ask before.’

Project details

Dates 2016–2018; 2017–2018
Funders The Marsden Fund; James Cook Research Fellowship, Royal Society Te Apārangi
Websites Multivariate statistics for ecologistsNZ Institute for advanced study