With fewer birds seen on farms, scientists try listening for them

Growing up, Adam Dixon couldn’t swim or fish in the Arkansas River which flowed through his town of Derby in Kansas.

Agricultural runoff, along with industrial activities, heavily impacted the river, often prompting authorities to warn residents to not eat fish from the water body. Growing up witnessing this issue informed Dixon’s career choices.

“I realized that agriculture impacts environmental quality, the quality of my life and the people in my community, as well as the biodiversity around us,” Dixon, now a conservation scientist at WWF, told Mongabay in a video interview.

That interest followed him to school. In 2015, as part of his Ph.D. at the University of Maryland, Baltimore County, Dixon started research into how to monitor birds in agricultural landscapes. He wanted to see how birds were faring on working lands, and how to gather data to help better manage vegetation to increase biodiversity. The initial stages of his research coincided with what he called “an explosion of technology” being used for conservation purposes. Encouraged, Dixon started thinking of tech solutions for his research.

“All the technology was converging and it seemed like a pertinent thing to try out,” he said. “I put together a low-cost experiment with the goal of increasing the total number of observations of birds and their habitats in croplands.”

In a study he co-authored and published earlier this year in the journal Ecological Applications, Dixon described how he used high-resolution satellite images to map croplands and then combined that with audio data obtained through passive acoustic monitoring to identify birds that live there.

“It boils down to this: Why do we see the types of birds we see in the farms and what habitats do they prefer?” Dixon said. “We were able to analyze biological information in the sounds, and then look for a statistical relationship between the habitat and the bird sounds.”

When it comes to biodiversity monitoring and conservation, farmlands often receive much less attention than forests and other protected areas. It’s not without reason. Logistical hurdles abound when it comes to accessing private farms and convincing owners to collaborate.

However, the numbers suggest that an increase in mechanized agriculture and the subsequent loss of habitat have impacted biodiversity on farmlands over the years. For example, farmland birds have seen a precipitous decline in countries in the Global North. According to the European Commission, the population of common farmland birds declined by 17% between 2011 and 2018, a far steeper rate than the 4% decline in the population of common bird species overall during the same period. A study published in 2018 in the journal Animal Conservation emphasized the need for urgent, large-scale interventions to protect and conserve farmland birds.

Dixon said his goal was to overcome those gaps and establish a blueprint for a feasible method to study birds on working lands.

His research took place across 44 agricultural sites in the state of Iowa over the course of 2019, and looked at non-crop vegetation that grew in the gaps between rows of crops, as well as at the edges of fields.

The first step was to map the locations. Using high-resolution satellite images, Dixon and his teammates determined the area of each pocket along with the variety and density of vegetation there. “More plant species and more structural variation has been linked to bird diversity,” Dixon said.

With the maps in hand, his next step was to monitor and identify birds that live in those habitats. For this, Dixon deployed audio recorders, some on his own and some with the help of farmers and landowners. Once the audio recordings — seven to eight days’ worth of audio — came through, he listened to them to spot birds and determine the species richness in each of the locations. “I was manually listening to morning calls and counting the species I heard,” he said. “I also had a graphical representation of the sounds which would show the highs and lows in the audio, and that helped me out as well.”

Using this method, Dixon identified 51 species of birds across all the sites, and as many as 26 vocalizing species at some locations. By combining the audio data with the satellite imagery, Dixon found that for every 10% increase in the area of land with non-crop vegetation, the species richness increased by 8%. The texture — variety and density of the vegetation — was, however, not an effective factor in determining species richness. Dixon said this could possibly have been because of the smaller sampling size, something he said he hopes will be corrected in future studies.

“I did expect the texture to help us predict birds, which it didn’t,” he said. “But I don’t think that necessarily made the idea null. What it did do was show that we need to go back and try other methods.”

Dixon said his pilot project could be adapted and further developed to serve as a model to monitor and study wildlife in croplands. To better understand the habitats, Dixon said data on other attributes, including biomass and dimensions of the vegetation, could be included.

Additionally, he said he hoped to use automated methods to analyze the audio data to process more data quickly and efficiently.

“AI has exploded to a point where we have a lot of automatic species identifiers,” he said. “That will make the analysis process even faster because we can set up a machine-learning algorithm to pick out all the different sounds.”

Banner image: Eastern bluebirds (Sialia sialis) in Texas, U.S. Eastern bluebirds are commonly seen in agricultural lands across North America. Image by Shiva Shenoy via Flickr (CC BY 2.0).

Abhishyant Kidangoor is a staff writer at Mongabay. Find him on Twitter @AbhishyantPK.

Podcast: A bittersweet bioacoustics bonanza

Citations:

Dixon, A. P., Baker, M. E., & Ellis, E. C. (2023). Passive monitoring of avian habitat on working lands. Ecological Applications, 33(5). doi:

Vickery, J. A., & Tayleur, C. (2018). Stemming the decline of farmland birds: The need for interventions and evaluations at a large scale. Animal Conservation, 21(3), 195-196. doi: