As Mitch Weegman was completing his undergraduate degree at Mississippi State University, a handful of professors encouraged him to apply for a Rhodes Scholarship, a prestigious award for students to attended graduate school at the University of Oxford in England.
Weegman had already received a Goldwater Scholarship for undergraduate research and decided he would throw his hat in the ring for a Rhodes Scholarship – he just needed to find a research project. Weegman started the process by sending emails to conservation organizations in the United Kingdom. One, Wildfowl and Wetlands Trust, responded.
“Wildfowl and Wetlands Trust is like Ducks Unlimited here in the United States,” Weegman said. “They said that they were interested in a collaboration with Oxford. They were working on a number of projects in waterfowl ecology, but focused on swans, scoters and white-fronted geese at the time.”
Weegman, an assistant professor in the School of Natural Resources, recalled seeing white-fronted geese often during their migration through North Dakota and South Dakota when he was growing up. Weegman had found his bird.
“I’ve always been interested in white-fronted geese,” Weegman said. “As I started researching them in greater depth, they became really compelling.”
The next step was writing a proposal for the Rhodes Scholarship, which Weegman completed and submitted. Unfortunately, Weegman didn’t end up with the scholarship – but another opportunity presented itself.
“Wildfowl and Wetlands Trust really liked my proposal,” Weegman said. “They decided to provide the majority of the funding in the collaboration with Professor Stuart Bearhop, and I was off to the University of Exeter to do my Ph.D. work.”
In 2014, Weegman completed his Ph.D., which focused specifically on Greenland white-fronted geese. This enigmatic population has been declining rapidly during the past 20 years, to the extent of about 1,000 geese per year. There were approximately 35,000 Greenland white-fronted geese globally in 1999. That number has dwindled to about 18,000 currently.
“We designed my original work to understand factors limiting this population,” Weegman said. “It didn’t make sense that this population was declining when a lot of waterfowl populations in Western Europe and North America were increasing. It was puzzling.”
Weegman’s original work is now being expanded through a collaboration with Texas A&M University-Kingsville and the University of Missouri. Along with Weegman, Stephanie Cunningham, a master’s student working with Weegman; Chris Wikle, professor in the Department of Statistics, who is working with Weegman to expand on Cunningham’s work; Toryn Schafer, a Ph.D. student working with Wikle; Bart Ballard, professor and research scientist with Texas A&M University-Kingsville; and Jay VonBank, a Ph.D. student at Texas A&M University-Kingsville working with Ballard, are collaborating to compare the decisions Greenland white-fronted geese make with the decisions North American white-fronted geese make.
The team is tracking geese, in North America and Western Europe, with fitted-to-neck collars to learn more about the decisions geese make during their migration. The team will be studying how those decisions affect population numbers, if at all. The goal is to gain a deeper understanding of the geese in both locations, with insights into how decision making explains population change in other genetically similar species.
“Are the two populations on different trajectories because of individual behavioral differences?” Weegman said. “Or are the populations differences a simple facet of different life histories, given the Greenland white-fronted geese and North American greater white-fronted geese are almost genetically identical? Is it something more cryptic than that? Do the Greenland birds manage their year incorrectly? It may not be as simple as the life history differences.
“If the North American birds leave Kansas and arrive to a blizzard in Nebraska, the birds are likely to return to Kansas. We have no evidence of the Greenland birds turning around and returning after initiating a migration episode. They endure some dangerous weather during migration. From previous tracking data, when the Greenland white-fronted geese hit bad weather, they continue to fly. The average route between breeding areas in Greenland and staging areas in Iceland, and staging areas in Iceland and wintering areas in Great Britain and Ireland, requires 17 hours of nonstop flight. In these crazy storms, we’ve seen 30- to 35-hour flights. It’s incredible. They’re averaging 10 to 15 miles an hour for 30 to 35 hours. That’s what these data allow us to see.”
Weegman and Ballard met two years ago at the North American Duck Symposium. As the two started to discuss past research, they realized their interests were incredibly similar.
“We realized that we had a lot in common,” Ballard said. “Our North American greater white-fronted goose project was just underway, and Mitch had some experience with the tracking devices we had just started using. I think we visited until 3 a.m. about research opportunities.
“Collaboration on projects like this are critical. Each researcher brings in different expertise and together it greatly strengthens the project. Dr. Weegman and his group have been exceptional to work with and have enabled the European aspect to be integrated into our North American project. The integration of information from greater white-fronted geese from both continents will allow us to investigate some questions that have not been able to be investigated previously, and that allows for a direct comparison between two populations of the same species that must cope with different challenges during migration.”
Ballard teaches and conducts research on the ecology and management of birds – with a focus on waterfowl. With funding provided by the Texas Parks and Wildlife Department, Ballard and VonBank had already began their research when Ballard and Weegman met. Their project is focused on investigating aspects of the movement ecology of white-fronted geese during migration and winter. These geese winter in Texas and Louisiana and breed in Alaska, with an entirely overland migration.
“Bird migration has always fascinated me, particularly when you consider the strategies that different species use, how these strategies evolved, and how they are successfully implemented,” Ballard said. “Geese are a great model to investigate migration because they migrate moderately long distances, employ a migration strategy in which decisions during late winter and spring can influence reproductive performance, and are large enough to carry an advanced tracking device.”
Tracking collars are being used on the geese, and Ballard and VonBank have put collars on 55 geese. The collars are being fitted to only adult, female geese.
“We are using tracking devices that provide locations every 30 minutes, and use the cell-phone network to automatically upload the information to a network,” Ballard said. “We can remotely download these data. We get GPS-quality locations and have on-board accelerometers that provide information on body movements that we can later analyze to understand behaviors. Thus, we get information every 30 minutes on where the geese are and every six minutes on what they are doing.”
Within the project, VonBank is researching waterfowl energetics, migration and wintering ecology in relation to nutrition, habitat relations and movement dynamics during migration and non-breeding periods. VonBank, working toward a Ph.D. degree in wildlife science, is researching questions addressing winter habitat use and selection, behaviors and energy expenditure, and movements and migration characteristics of North America greater white-fronted geese.
“My role is addressing several questions pertaining to migration and wintering ecology of greater white-fronted geese,” VonBank said. “Over the past two winters, I have been in charge of conducting captures of geese throughout the state of Texas to deploy our tracking devices, which provide data necessary to answer both my graduate degree questions at Texas A&M University-Kingsville and the Caesar Kleberg Wildlife Research Institute, as well as the questions we are asking collaboratively with Dr. Weegman and students at the University of Missouri.”
This isn’t the first time that VonBank has worked with waterfowl. He worked at a waterfowl research center in Illinois from 2013-15 as a research assistant. He also has helped other graduate students with research at the University of Illinois who are working with Canada geese.
“I am thrilled to have Dr. Weegman and students at the University of Missouri involved in this research,” VonBank said. “It allows us to parse out specific questions among individuals who take the lead, but the process benefits from the expertise and input from the entire group. All of the work collectively paints a very extensive and thorough picture of processes throughout the annual cycle of greater white-fronted geese that are understudied or completely unknown.”
Greenland White-Fronted Geese
Greenland white-fronted geese breed in Greenland, stage in Iceland and winter in Ireland and Scotland. The migration is treacherous, as the birds fly nearly 1,800 miles, with the majority of that flight taking place over the North Atlantic Ocean. That means there are really no stopping points or energy sources for the birds during the flight. These geese sometimes fly for 30 hours straight.
Their population is counted in Ireland and Scotland, and the number of young are also surveyed during that time. Past research had shown a decrease in population numbers – and Weegman’s Ph.D. work looked at why that decrease was occurring. He fitted some of geese with backpack tracking devices, which were used to understand whether individual birds were attempting to breed and failing or just not attempting to breed at all.
“It was this initial work that suggested that very few individuals were even attempting to nest or that those that attempted had failed early during incubation of eggs,” Weegman said. “We had a very small sample size, with just 15 birds. However, only two of those were successful and brought back young to wintering areas.
“That’s not a sustainable strategy for animal populations.”
Weegman’s research showed that the climate in Greenland has played a role in the decrease of population. Thirty years ago, seven of every 10 years produced low snowfall. Currently, seven of every 10 years are producing serious snowfall.
“We are seeing increased sea surface temperatures off of the west coast of Greenland,” Weegman said. “Increased sea surface temperatures mean increased precipitation in west Greenland, and increased precipitation in west Greenland in May means snow. We know from arrival dates that the geese appear to be arriving in Greenland at the same time that they did 30 years ago. They’re arriving to snow now. That is problematic because the geese need to reacquire nutrients prior to laying a clutch of eggs, but can’t access food underneath snow.”
Building on the Past
About one year after Weegman and Ballard met, they traveled to Iceland to begin new research. During that first trip, the group put collars on 10 geese. The collars are the same as the ones used by Ballard and VonBank for the geese in North America.
Since then, Weegman, Cunningham, Schafer and VonBank deployed collars on 11 geese in Ireland in March 2017 and 19 geese in Iceland in September 2017. The group also filmed the geese to get a better understanding of their movements.
“My first task will be to watch the film that we recorded of individuals fitted with the GPS collars,” Cunningham said. “We need to ground truth the accelerometer data. We’ll line up the video with the acceleration signal, and then be able to assign the specific behavior the individual is performing to that signal. We’ll collect as many labeled segments of accelerometer data as we can, and that’s what we’ll use to train the algorithm.
Cunningham is looking at the decision making of the two populations as part of her master’s research.
“Once I’ve compiled a training dataset, I’ll use machine learning algorithms to classify the remaining data, encompassing time periods when we weren’t able to directly observe the geese,” she said. “I’m going to look at the behavior, mostly during spring migration and through the breeding season, and the accelerometer data will allow me to establish activity budgets. The birds in North America and the birds in Greenland have different migration routes. The North American birds are going through the central United States and Canada, where they can stop frequently in places like Nebraska and take advantage of the agricultural landscape for food. The Greenland birds are crossing the North Atlantic, and so their options for stopping to refuel are limited. I’m also going factor in environmental effects, such as storm events, and see how they impact behavior and the decision to attempt breeding. It’s going to be interesting to compare the two populations.”
Cunningham is working with Schafer, a Ph.D. student in statistics, who is putting together statistical modeling based on the data.
“My work is a lot of modeling,” Schafer said. “It’s parallel to what Stephanie is working on with behaviors. Currently, I’m analyzing the sequences of behaviors that Stephanie is going to classify. Once she does that, we’ll have a sequence of behaviors every six minutes, from sitting, flying to walking.
“I am working on predicting the probabilities of being in those behavioral states, as well as switching to other behavioral states. It’s not the same day-to-day and location-to-location. We’re looking to build those components into our model.”
Cunningham had never worked with waterfowl before starting this project. She had, however, done some work with accelerometers.
“I wasn’t looking to work with one particular species when I was applying for grad schools,” she said. “I saw Dr. Weegman’s project advertised, and even though I’m not from a waterfowl background, I had used accelerometers before. That’s what drew me in. Before I started here, I spent six months at the Smithsonian Conservation Biology Institute and eight months in Texas at Fossil Rim Wildlife Center – the whole year was focused on work with scimitar-horned oryx, a large antelope that went extinct in the wild in the 1980s. In Aug. 2016, a collaboration of several international organizations, began releasing groups of individuals back into their native habitat.”
Before the oryx were released, several captive individuals were fitted with GPS collars – which contained accelerometers. Cunningham, who received her bachelor’s from Michigan State University, was in charge of charting the daily behaviors of the oryx. The goal was to see if the collars had any impact on behavior, and the accelerometer data was used to supplement visual observations.
After working with large mammals, Cunningham has shifted her focus to geese. She is working to determine whether successful breeders are showing specific behavioral patterns.
“The use of accelerometers in wildlife research is a relatively recent development,” Cunningham said. “I’m excited to be able to utilize this technology in a real-world situation to help answer important questions related to why these birds show such disparities in population trajectories.”
“We filmed these individuals with devices in the wild,” Weegman added. “We know, based on the film, the behavior of the individual. We can then assign thresholds. For example, feeding might be one to two meters per second. Flying is above two meters per second. Sleeping is less than one meter per second. We can then use these machine learning algorithms categorize the entire data set of acceleration values. Every six minutes is a remarkably frequent interval to understand wildlife behavior throughout the year.”
Schafer, who received a bachelor’s in wildlife biology and statistics at Colorado State University, had never worked with waterfowl before this project, either. She has always had a passion for ecology, and is self-funded through the National Science Foundation.
“I had some flexibility with what I could work on, and it’s exciting to be a part of this project,” she said. “It’s hard to classify a large suite of known behaviors accurately. We know that animals have a lot of behaviors. There’s a trade off between being able to label the behaviors and being able to have a rich class of states. We’ll compare the results from assuming a small, known set of behaviors to assuming we only observe the raw acceleration data. The latter set up asks, given the acceleration data, can we uncover the hidden true states that they’re in?”
Data is constantly coming in on all 95 geese that have collars on them. The team will dig through the data as more of it filters in. The University of Missouri College of Agriculture, Food and Natural Resources is funding Weegman’s part of the project.
“It’s a really cool collaboration,” Weegman said. “We’re excited to see where the data takes us.”