Research

My research uses large biodiversity, environmental and satellite datasets to ask where species are, how they are changing, and whether protection and management are keeping up. It runs from New Zealand rivers and coastlines to the global ocean, and almost all of it is built as reproducible code.

13peer-reviewed papers
5as first author
1more as a preprint
5scientific reports
Full publication list Google Scholar Academic CV
Spatial conservation prioritisationReinforcement learningRandom forests and gradient forestsSpecies distribution and density modelsNull and permutation modelsMixed effects and quantile regressionFunctional and phylogenetic diversitySatellite and vessel-tracking dataR, Python and Quarto

In each list, a dark bar marks a first-author paper.

Global map of standardised effect sizes of species range coverage by no-take marine protected areas
Gap analysis: how well no-take protected areas cover species ranges compared with a null model of random placement (blue better, red worse than chance).
Percentage of range protected by no-take MPAs for each IUCN Red List category
Share of each species' range inside no-take protected areas, by IUCN Red List category: threatened species are no better covered.
Bivariate map of conservation priority against industrial fishing effort
Conservation priorities from reinforcement learning, set against industrial fishing effort.

Global macroecology

Area-based management and conservation prioritisation

How well does the global network of marine protected areas cover biodiversity, and where should new protection go? With the Pimiento Research Group, I combined range maps for about 1,000 species with measures of extinction risk, ecological function and evolutionary history. I ran a gap analysis against a spatially explicit null model, then used reinforcement learning to optimise where a 10% protection target would do the most good, set against where industrial fishing concentrates.

The study system is sharks and rays, one of the most threatened groups in the ocean; the methods apply to any taxon or region.

Cumulative number and area of marine protected areas by IUCN category, 1940 to 2023
Cumulative number (A) and area (B) of marine protected areas by IUCN category in the Central and South American Pacific, 1940 to 2023.
Lollipop charts of EEZ coverage by no-take and all MPAs for 11 countries
Share of each country's exclusive economic zone under protection: expansion since 2010 (A) and extent in 2023 (B), no-take against all MPAs.

Science to policy

Protection gaps in critical habitat

As scientific coordinator for the IUCN SSC Shark Specialist Group, I worked with more than 60 experts to identify Important Shark and Ray Areas, then tested how well existing protected areas actually cover them. The answer, region after region, is very little, especially once fishing is taken into account.

Animated map of Chinese industrial fishing effort accumulating through the year
A year of Chinese industrial fishing effort accumulating week by week, from AIS data (frame from the animation).
Grid of marine protected areas by country coloured by detected oil-slick area
Oil slicks inside marine protected areas, 2023 to 2025: each square is one MPA, coloured by the oil detected inside it, for the five most exposed countries.

Open analyses

Human pressure at sea, seen from space

Much of my recent work uses public satellite data to measure human activity at sea: industrial fishing from vessel tracking (AIS) and Sentinel-1 radar, and oil pollution from radar slick detections. These are open, reproducible analyses, each with its code.

Temporal trends in taxonomic and functional diversity of river macroinvertebrates
Taxonomic homogenisation and functional differentiation in New Zealand rivers.
Map of New Zealand river monitoring sites
The national river monitoring sites behind the Global Change Biology study.

Global change

Biodiversity change under climate and land use

My PhD used three decades of national monitoring data to separate the effects of climate and land use on freshwater communities, showing that species composition and ecological function can respond in opposite directions. Related work tracks how nutrient enrichment reshapes coastal lagoon communities.

Spatial mismatch between taxonomic, functional and phylogenetic diversity of New Zealand cetaceans
Mismatch between diversity facets of New Zealand cetaceans.
Modelled probability of occurrence of a cetacean species in New Zealand waters with zoomed insets
Modelled probability of occurrence of a cetacean species across New Zealand's waters, with zoomed insets (Stephenson et al. 2020).

National scale

Mapping and protecting marine life in New Zealand

At NIWA I modelled where New Zealand's whales and dolphins live and classified its seafloor communities with machine learning, work that fed conservation planning for Fisheries New Zealand and the Department of Conservation.