Rigorous answers to questions that matter.

Fishers (Pekania pennanti) remain restricted to an isolated population in southern Oregon, and why they haven't expanded out despite legal protection remains a mystery. They share their forests with cougars and bobcats, larger carnivores that compete with them for prey and kill them, but cougars may also facilitate fishers, indirectly provisioning ungulate carcasses that fishers scavenge. Whether direct or indirect interactions between these species shape where fishers can persist, and how that should inform management and future reintroductions, is an open question.

Working with the Oregon Department of Fish and Wildlife, we capture fishers, cougars and bobcats each winter and fit them with GPS collars and daily diary tags, tri-axial accelerometer and magnetometer loggers that record fine-scale behaviour second by second. A grid of remote cameras, analysed with spatial mark–resight models, estimates how many carnivores share the landscape. Together with diet analysis, these data build a mechanistic picture of how the three carnivores divide up space, prey and risk.

The coastal marten (Martes caurina humboldtensis), the coastal subspecies of the Pacific marten (Martes caurina) also known as the Humboldt marten, survives in a handful of small, isolated populations in coastal Oregon and northern California, and is federally listed as threatened. Recovery planning depends on numbers that are very hard to get: how many martens there are, how well they survive, and how many young join the population each year.

On the Oregon coast, including the shore-pine forests of the Oregon Dunes, scent-detection dog teams search for marten scats, and DNA from those scats identifies individual animals without anyone needing to catch one. Spatial capture–recapture models turn those genetic recaptures into estimates of density and vital rates, the quantities that tell managers whether a population is holding on.

Monitoring only helps a threatened species if methods can detect the species when present, and also reliably detect changes in population dynamics. Based in northern California, this project combines remote-camera surveys with GPS tracking of martens to estimate density, and then uses simulation-based power analyses to ask how many cameras, how far apart and for how long a monitoring programme needs to run to reliably be able to detect a marten given it is present. The goal is a sampling protocol that agencies and forest managers can apply across the coastal marten's range.

Tessa also leads our test of airborne environmental DNA, sampling the air itself for traces of the mammals that live in a forest, against remote cameras in OSU's McDonald–Dunn Research Forest.

The gray-tailed vole (Microtus canicaudus) lives almost nowhere but the Willamette Valley, where its populations periodically erupt. In outbreak years voles strip grass-seed, clover and other crops across the valley. Between outbreaks they can be hard to find at all. Why populations boom, why they crash, and whether an outbreak can be forecast are open questions with real costs for growers and for the predators that depend on voles.

We are building long-term monitoring of vole populations, using live-trapping and open spatial capture–recapture models to follow density, survival and recruitment through the cycle. Alongside it, we screen voles for pathogens to test the role of disease, and fit miniature biologging tags to measure how voles move and behave.

Managing a rodent that damages crops is as much a social problem as an ecological one. We are part of the Willamette Vole Management working group, which brings growers, researchers, agencies and non-governmental organisations together to co-develop science-based, grower-led strategies for living with voles.

With the group, we help organise the annual Willamette Valley Vole Summit, where growers and researchers compare notes and map where and when voles have hit hardest. We are also building large experimental enclosures at OSU's Hyslop Farm, where vole densities can be set and held, so that the damage a given number of voles does to a grass-seed crop can be measured directly rather than guessed.

The western gray squirrel (Sciurus griseus) has declined across much of the Pacific Northwest and is listed as state-endangered in Washington and a species of greatest conservation need in Oregon, yet there is no established way to count it over large areas. We are developing one.

Hair-sampling stations collect DNA without capturing animals, individuals are identified with a purpose-built genetic marker panel, and spatial capture–recapture models convert those detections into density and demographics. Camera-based occupancy surveys extend the picture across the wider landscape, and simulation-based power analyses show how much survey effort is needed to accurately estimate population sizes and identify drivers of decline.

Large wildfires are reshaping western Oregon's forests, and the decisions that follow a fire are often made with little information about how wildlife responds. In burn scars outside Roseburg, we survey birds with autonomous recording units and point counts, and mammals with remote cameras.

Bird data from the two methods are combined in an integrated distance-sampling framework to estimate density, and mammal detections feed multispecies occupancy models. Together they show how whole communities respond to fire and to the land management that comes after it.

Much of our methodological work puts a blunt question to the models ecologists rely on: do they recover the truth when we know what the truth is? Through simulation and case studies we have shown when co-occurrence models can, and cannot, reveal interactions between species; developed abundance-mediated interaction models that let one species' numbers, not just its presence, affect another; combined opportunistic records with structured surveys for species that are difficult to monitor; and built spatially explicit power analyses for multi-species monitoring programmes.

We are currently developing hierarchical distance sampling for camera-trap data, a way to estimate the density of animals that cannot be individually recognised from how far away cameras detect them. Students in the lab meet these tools in FW 599 Applied Hierarchical Modeling and put them to work in every project above.