Combining Multiple Survey Data Strengthens Monitoring Efforts of Imperiled Game Birds

Bird populations have steeply declined across North America over the last century, and those that typically dwell within the grasslands and prairies have faced the greatest challenges. This is primarily the result of the destruction and conversion of these native habitats to urban environments and intensive agriculture. In addition, an essential part of the diversity and long-term success of native grasslands are periodic fires, which have been interrupted by landscape change, allowing the encroachment of woody growth and invasion of non-native, aggressive plant species. Because of these combined factors, more than 60 percent of North America’s native grasslands have disappeared since the pre-industrial era. By 1978, less than 2,300 acres of Illinois’ original 22 million acres of prairie remained. Consequently, it is estimated that grassland bird populations have declined by more than half since 1970, with some species seeing even greater reductions in their numbers.

Two grassland bird species which have declined in previous decades are the northern bobwhite quail (Colinus virginianus) and the ring-necked pheasant (Phasianus colchicus). These declines are largely attributed to the loss and fragmentation of their preferred habitat, including within Illinois. Ring-necked pheasants are not native to North America, previously introduced from Asia for sport, but both species are now important game species of significant cultural and economic importance in the USA.

Bobwhites prefer landscapes with a mixed variety of habitats, with plenty of ‘habitat edges’ (the interface between two or more habitat types), and often use woody cover. They thrive in weedy meadows, open forests, and clear cuts with sufficient native grasses. Native warm-season bunch grasses are important to bobwhites, as they not only use this vegetation to avoid danger, but also use them as nesting sites, hiding their eggs from the view of predators. These grasses, when properly managed, also provide room at ground level for these birds and their chicks to freely move to forage, while providing overhead cover for concealment.

Ring-necked pheasants primarily rely on grasslands but can be reliant on other habitats for some parts of the year. Nesting season in early spring and summer generally requires at least moderately dense grasslands, which has historically included unmowed roadsides, while the brood-rearing period after hatching requires more open structure at the ground level for chicks to navigate. In the fall and winter, pheasants make greater use of farm fields to eat grain and can be found using densely vegetated wetlands and windbreaks in addition to dense grasslands. Similar to bobwhites, pheasants are also ground-nesting birds, nesting among tall vegetation to hide their eggs from predators.

Due to the loss and alteration of natural grasslands and shrublands it is important to monitor populations of birds that are dependent on these habitats to ensure their continued persistence. Beyond these species-specific conservation goals, birds can also act as biological indicators, signaling to biologists any changes in habitat quality which may also be impacting other priority wildlife species.

Biologists measure the status of bird populations in different ways, and each approach may have different goals. Monitoring programs are often targeted toward a particular species or habitat. Although this focused approach may increase efficiency for sampling a given species and its habitat, failing to collect information from all potential habitats may lead to inaccurate estimates. Another potentially powerful type of data for determining the status and trends of bird populations relies on observations collected by the public. One such data source, The Cornell Lab of Ornithology’s eBird, involves the public with ecological research and potentially engages people in the management and conservation of birds. Citizen-science data, such as eBird, however, have limitations like all others. One particular concern is that these data are collected only where people want to go, meaning a tendency to unintentionally favor collection of data from some areas over others. For instance, in Illinois, the majority of eBird data are recorded within the Chicago Metropolitan area, where most people reside, but is not necessarily where these birds prefer. Another source of volunteer-based bird data, the North American Breeding Bird Survey, is designed to avoid spatial bias by randomly locating sampling routes, but this approach may be less efficient at detecting habitat specialists such as game birds.

Population Monitoring of Imperiled Grassland Birds

So how do we utilize the strengths and mitigate the weaknesses of each monitoring design if we want to provide the best population monitoring of imperiled grassland birds?

In recently published research in Ornithological Applications, Dr. Robbie Emmet and fellow biologists at the Illinois Natural History Survey combined data from multiple bird surveys conducted within Illinois. They created a new statistical model that combined data from The North American Breeding Bird Survey, eBird, and targeted upland game counts conducted by the Illinois Department of Natural Resources, to predict the effects of landscape composition on the occurrence of northern bobwhites and ring-necked pheasants throughout the state. They compared statistical models composed of each combination of survey datasets to predict occurrences of both bird species. This allowed the researchers to determine whether combining survey data from different sources improved the ability to predict occurrence of these species.

Although the point of the exercise was to explore the potential advantages of combining data sources rather than determining habitat associations for bobwhites and pheasants, a topic that’s already well explored, the new approach did confirm these known associations. More specifically, bobwhite occurrence was strongly associated with edges and pheasants were less likely to be found in forested landscapes, but more importantly Dr. Emmet and his team were then able to use these landscape associations to predict game bird occurrences throughout the state.

How Did the Different Survey Data Improve the Model?

The authors found that using combined datasets, in this case two or three sources, led to better predictions. Since each respective dataset had different strengths and weaknesses, this meant that combining them compensated for the shortfalls in any given data type. For example, the upland data from the Illinois Department of Natural Resources was important for making accurate predictions while the Breeding Bird Survey data helped improve the certainty of those predictions.

Altogether, incorporating multiple data sources increased the performance of the occurrence models for both bobwhites and pheasants. This may have important implications for the management of our game birds in Illinois, showing that targeted surveys provide the highest-quality data, but less targeted, more widespread surveys can improve precision and account for those biases we mentioned earlier. Even though data from surveys such as eBird have some biases when used alone, they can help strengthen targeted surveys, filling in the ‘blind spots’ by the targeted sampling design.

Dr. Emmet commented that these results “[demonstrate] the enormous contribution of community scientists to research.”

Based on the success of this and similar studies, Dr. Emmet would encourage wildlife managers to combine multiple data sources, when possible, for monitoring other vulnerable wildlife populations.

---

Nathan Proudman is a Postdoctoral Research Associate at the Illinois Natural History Survey. His research has primarily focused on the ecology of mammals. Currently, he is working on a statewide monitoring program for mammals in Illinois.