The method Montana’s wildlife agency uses to estimate wolf populations may be so flawed that it’s over-estimating the wild canine’s numbers by 150%, a Bozeman wildlife researcher wrote in a recent study.
However, six state and federal researchers countered the new study has flaws "that should strongly call into question the merits" of the research, which they said was "weakly supported" and "must be considered preliminary."
At last count, Montana Fish, Wildlife & Parks estimated the state’s wolf population at 1,087 animals in 181 packs in the western third of the state. The state has been using the estimates since 2020.
“Clearly the number one criteria for, or metric of managing how much you have of something, like wolves or elk, is abundance,” said Robert Crabtree, founder and chief scientist of the private nonprofit Yellowstone Ecological Research Center, and lead author of the new wolf study. “So it’s like a rancher who has a massive rangeland, and he doesn’t know how many cows he has.”
People are also reading…
Crabtree is well-known for his controversial research decades ago showing that killing coyotes as a means of predator control can backfire, as the animals will produce more pups. He co-authored the recent study examining wolf counts with help from statisticians Dean Koch and Subhash Lele from the University of Alberta. The study was funded by the Jodar Family Foundation and the Rangeland Foundation.
Patch occupancy
To estimate wolf populations, Montana Fish, Wildlife & Parks uses an analytic tool called integrated patch occupancy model, iPOM for short. The model was developed with the help of Sarah Sells, from the Montana Cooperative Wildlife Research Unit at the University of Montana in Missoula.
“Patch occupancy models are statistical models that use detection/non-detection data to estimate the probability of a species’ occurrence over a specific area while accounting for the fact that a species is rarely detected perfectly,” wrote Sarah Bassing, a University of Montana wildlife biology grad student.
“Conceptually, iPOM is a good idea,” Crabtree said in a phone interview. “They just did it all wrong.”
To ensure the researchers were properly duplicating FWP’s methodology, Crabtree requested the code Sells used for iPOM, as well as her data and models and read her research. When questions arose, the researchers talked to Sells on the phone this past spring. In an attempt to be transparent and objective, Crabtree notified Sells of the findings so she could comment on the study as it has gone public for preprint review before submission to academic journals for peer review.
"We believe that peer review will address the severe misinterpretations in the Crabtree et al. analysis prior to being published, and if it is published in the peer-reviewed literature we will respond in detail in that forum," Sells wrote in a response to Crabtree's research requested by The Billings Gazette.
The two-page response was also signed by FWP officials Justin Gude, Kevin Podruzny, Molly Parks and Brian Wakeling. Michael Mitchell, a former colleague of Sells at the Montana Cooperative Wildlife Research Unit who is now retired, also signed the document.
Wolf plan out for comment
Crabtree's study, titled “Misleading Overestimation Bias in Methods to Estimate Wolf Abundance that use Spatial Models,” is undergoing scrutiny as Montana Fish, Wildlife & Parks is taking public comment on its draft wolf management plan through Dec. 19.
Crabtree’s study is also the latest negative review from a string of noteworthy scientists who have voiced concerns about FWP’s wolf counting methodology, including Montana State University wildlife researcher Scott Creel, retired FWP wolf and carnivore specialist Diane Boyd and retired Yellowstone National Park Wolf Project biologist Doug Smith.
“I think the issue is using a distribution estimator for abundance — which is possible but very difficult,” Smith wrote in an email when asked about Crabtree's research. “Crabtree et al. points out just one assumption violation, area of occupancy. There are likely others, which raises the question is there a better method or solution for estimating wolf numbers?”
Criticisms of iPOM
One of Crabtree’s criticisms of iPOM is its use of large grids when assessing the total area occupied. In Montana, the model uses 600 kilometers squared.
“We found that even the use of small grid cells (relative to wolf territory size) inflated the total area occupied (AO) by wolf territories as well as the number of wolf packs,” the researchers wrote. “Furthermore, the overestimation bias rapidly increases with increasing grid cell size.
“Given the sensitivity of how iPOM’s spatial models result in severe overestimation bias, we fail to see how iPOM could detect any change in abundance except possibly at or near extirpation levels,” Crabtree and his colleagues wrote.
Another potential problem the paper cited was iPOM’s use of hunter observations of wolves in the fall.
“iPOM appears to use problematic definitions of how hunter observations are used to determine pack versus non-pack members during the fall when already highly mobile wolves undergo hunting mortality that affects pack composition, dispersal, and territorial status,” the researchers wrote.
Consequently, the scientists argued fall is the least appropriate time to collect such data.
Research and development
Because counting wolves is so difficult, given their roaming nature and the rugged terrain they occupy, FWP worked for six years to develop its model for estimating wolf populations. In 2007 the agency began using patch occupancy model, or POM for short, as well as manually counting the animals.
After noting differences between POM and field observations — mainly that pack sizes were being underestimated — FWP and Sells worked to improve POM, which resulted in the development of iPOM. In addition, FWP continues to use direct observation of wolf pack territory and size to verify the reliability of iPOM.
The changes haven’t removed the problems, Crabtree and his fellow researchers claimed.
“Given the current design of POM and iPOM, there is no ability to detect change let alone determine wolf population size,” the study’s authors wrote.
The researchers have posted their study online in order to be transparent and to allow other scientists to repeat their assessments to improve how to reliably estimate wolf abundance. Following that, their analysis of iPOM will be submitted to a scientific journal for peer review.
FWP response
In her response to the Gazette, Sells explained iPOM uses three submodels: an occupancy model, a territory size model, and a group size model. She said Crabtree primarily focuses on the occupancy and territory models to reach his conclusions, based on a flawed interpretation: "that the entirety of any (600 kilometer squared) grid cell with wolf observations is included in the sum of area occupied."
In iPOM, Sells said the total area occupied by wolves and the resulting estimates of number of packs and wolves is drastically reduced compared to the method Crabtree used. She then gave an example, saying the peak population size in 2011 under Crabtree's methods would estimate 402 packs and 2,310 wolves, whereas iPOM estimated 188 packs and 1,259 wolves.
Were Crabtree's calculations correct regarding the "true abundance estimate for wolves by a factor of 2.5," Montana's wolf population would be "well below the minimum counts" by FWP staff "of verified packs and wolves known to exist in Montana each year," Sells wrote. Although such field counts don't represent the total wolf population, "we are absolutely confident that more wolves exist on the ground than are represented by these counts," she added.
Sells' study has been peer reviewed, and she said the same process would identify and correct Crabtree's "flawed interpretations" and "myriad other unsupported assumptions."
"Technicalities of complex abundance estimators, their properties, and their accuracy are best discussed under this process, where errors can be preempted or debated with rigorous oversight by experts," Sells and her colleagues wrote. "Bypassing this route can lead to misinterpretation, misunderstanding, and misleading inferences that confuse rather than contribute to effective public debate."
After seeing Sells and her colleague's comments, Crabtree rebutted that his team used FWP's models, software code and data.
"So FWP can say whatever they want and say that we misinterpreted them, but again it was their data, their methods, their models and their software," he wrote in an email. "They bear the burden of proof to demonstrate what they are saying by doing the simulations and the testing that we did. Until then it's just all baseless words."
Judge weighed in
iPOM was raised as an issue in a 2022 court ruling. In dissolving a temporary restraining order on wolf hunting and trapping, District Court Judge Christopher Abbott found that even if iPOM were flawed, it was unlikely to result in Montana’s wolf population dropping below the 150 animals required to meet federal recovery goals after the animals were removed from Endangered Species Act protection in 2011.
Abbott also wrote: “The underpinnings of the iPOM have been peer-critiqued. Research on iPOM has generated two doctoral dissertations, has been presented to the public in numerous ways, and has been modified based on feedback from the scientific community, stakeholders and the community.”
Last hunting and trapping season, 258 wolves were killed in Montana, although FWP had set a quota of 450. This hunting and trapping season, which extends to March 15, the Fish and Wildlife Commission set the wolf quota at 313 wolves, six of which can be killed in a hunting district near the northern border of Yellowstone National Park. Conservation and tourism groups have opposed the killing of wolves so close to the park, since many of the animals spend most of their time in Yellowstone.
As of Dec. 12, FWP reported four wolves have been killed in the Yellowstone-adjacent hunting district with another 78 killed elsewhere in the state.
The value of wolf watching in Yellowstone National Park has been estimated at $82 million, a portion of which flows into Montana's gateway communities.
Wolf legislation, lawsuits
In 2021 the Montana Legislature passed a bill to increase the state’s wolf harvest by making more and cheaper licenses available, allowing snares, hunting at night and the use of bait on private land as well as payments for killing the animals. The Republican-dominated body also passed a law mandating a reduction in the state’s wolf population. Despite the moves, the wolf harvest has fallen the past two years. Last year’s decline was partially blamed on heavy snow that made trapping difficult.
In October, the Montana Fish and Wildlife Commission admitted to violating the constitutional rights of Wolves of the Rockies, an advocacy group, by not fulfilling its request for documents. Emails and phone records showed the commission had convened an illegal quorum in 2021, deliberating the outcome of wolf hunting and trapping regulations prior to a public meeting.
Wolf-watching value
Wolf advocates have long argued before the Montana Fish and Wildlife Commission, which has the final say on wolf hunting and trapping quotas, that there are fewer wolves on the landscape than FWP’s model estimates.
"To no one's surprise, another independent expert — Dr. Crabtree — has demonstrated that FWP's population model, iPOM, is fundamentally flawed and severely overestimates the state's wolf population,” said Lizzy Pennock, an attorney for WildEarth Guardians. “Importantly Dr. Crabtree's study also provides well-established and feasible alternative methods FWP could use instead, which we implore the agency to do."
Crabtree said since the study preprint has gone online he’s received numerous calls.
“I think this issue is going to blow up, and it kind of already has,” he said.