A Better Count of Blister Rust in Yellowstone’s Whitebark Pines

Park scientists who had counted blister rust in Yellowstone-area whitebark pines for years found that their own method undercounted it. The Greater Yellowstone Inventory and Monitoring Network reanalyzed its data with newer Bayesian hierarchical models, and the original approach underestimated both how common the disease is across the population and the chance that any single tree is infected.

Whitebark pine in the Greater Yellowstone Ecosystem (NPS/John Fothergill).

What is at stake

Whitebark pine (Pinus albicaulis) is a declining keystone species of high-elevation forests. In 2020 the U.S. Fish and Wildlife Service proposed listing it as threatened under the Endangered Species Act. The Interagency Whitebark Pine Monitoring Program, led by the Greater Yellowstone Inventory and Monitoring Network, has tracked the trees' health there since 2004 under a peer-reviewed protocol that several other agencies also use. A principal stressor is white pine blister rust (Cronartium ribicola).

The authors, NPS ecologist Erin Shanahan and colleagues, argue that a monitoring program has to stay adaptable. It should re-examine its statistics and field methods from time to time so that its results remain accurate and useful.

Why field monitoring is hard

Long-term field studies face problems a laboratory does not. Crews change often because the work is seasonal, the weather is unpredictable, and topography differs from site to site. Tree height, position in the stand and density vary too. Training helps, but some problems cannot be fully fixed.

Three observers debating evidence of white pine blister rust in the upper branches of a whitebark pine (NPS/Erin Shanahan).

Imperfect detection. Observers sometimes record a species or a feature as absent when it is present, a false negative. Wildlife studies have long handled this with occupancy models. Vegetation studies rarely do, on the assumption that plants do not move, yet a plant's size, coloring and position among other vegetation can make it hard to spot. For whitebark pine, the main measure was the status of blister rust infection, and rust is hard to see in the upper branches of tall trees and hard to assign to one tree when branches intertwine with neighbors.

Confounding factors. Another stressor, the native mountain pine beetle (Dendroctonus ponderosae), complicates the picture. An epidemic in the mid-2000s killed 70% of overstory whitebark pine. That shifted the population toward smaller trees, which are less likely to be infected than larger ones. The number of trees also varies from one belt transect to the next, so a transect with many trees pulls the infection estimate toward itself unless the analysis adjusts for it.

Mountain pine beetles killed the whitebark pine on the left, with red needles, in the early 2000s. The gray trunks in the center are whitebark pine that likely died in a smaller beetle outbreak in the 1950s on the Beaverhead-Deerlodge National Forest (NPS/John Fothergill).

What the study compared

The original protocol, designed around 2003, estimated infection with a design-based ratio estimator, a logical choice at the time. It did not account for imperfect detection, and it limited the exploration of factors that explained changes in infection over the 16 years of the study. The new work compared that estimator with model-based approaches that account for imperfect detection and for variation in infection probability across time and place.

In the most complete model, the infection covariates are tree DBH (a larger diameter means a higher probability of infection) and stand elevation. The detection covariates are tree DBH (infection in the canopy of larger trees is harder to detect), transect slope, transect hiking time and observer experience.

The original ratio estimator and four Bayesian hierarchical models, from the fewest factors to the most (NPS/Erin Shanahan).

The Bayesian hierarchical occupancy model gave the more complete picture. It was flexible about imperfect detection and inconsistencies in the sampling design, and it brought in explanatory variables for changes in infection over time.

What the authors recommend

• Use model-based approaches such as Bayesian hierarchical models for status and trend assessments of significant natural resources. Observational data carry uncertainty, and a model-based analysis accounts for the many sources of variation in a long study, which the authors see as key to management and policy decisions for at-risk whitebark pine.
• Re-evaluate protocols periodically. Checking models and diagnostics may show that the planned analysis can describe the data better, and better estimates improve what long-term monitoring says about status and trends.

Aecia, the spore-bearing structures of white pine blister rust, on the trunk of a whitebark pine. When mature, the spores erupt from blister-like pouches and are carried by the wind (NPS).

Why managers care

In the Greater Yellowstone Ecosystem, blister rust acts alongside chance events such as the recent beetle outbreak, other stand-altering events such as wildfire, and projected climate change. Analysis that can sort out these interacting effects shows patterns of infection and the possible consequences of the disease for future whitebark pines. Park managers then have current data for protecting and restoring the species on the landscape.

Sources

• Shanahan E., Wright W.J. and Irvine K.M., Adaptive monitoring in action: Reconsidering design-based estimators reveals underestimation of whitebark pine disease prevalence in the Greater Yellowstone Ecosystem, J Appl Ecol 2021, as summarized in the National Park Service article Advancements in Analytical Approaches Improve Whitebark Pine Monitoring Results by Erin Shanahan (NPS Inventory and Monitoring Division) and Kathryn M. Irvine (U.S. Geological Survey); rewritten in hubnx's own words. The paper's own figure is not reproduced here.