One of my interests since I began nature sound recording is how acoustic information can be used for scientific purposes. When an animal is recorded, a lot of information is saved. The identity of the species, the type of call, the date, the location. While each individual data point may not be that useful in itself, when accumulated over space and time, the recordings create an invaluable permanent record. More and more acoustic studies are taking place, from measuring noise levels in national parks, recording vocalizations of whales, quantifying changes in soundscapes over time. These recordings then become a valuable archive to document environmental change, climate change, recovery from a wildfire, impacts of logging and habitat loss.
About 15 miles south of where I live is a small Nature Conservancy preserve known as River Fork Ranch. It encompasses the confluence of the east and west forks of the Carson River, which arises in the eastern Sierra Nevada and ends at Stillwater National Wildlife Refuge near Fallon, NV. Where these two forks meet is a lovely wetland, surviving in spite of years of dividing the river into a series of sloughs to irrigate cattle pastures throughout the Carson Valley. The Nature Conservancy purchased this property in 2000, and has been actively restoring the function of the wetlands by restoring native plants and letting beavers reestablish. The property is an active cattle ranch and they work closely with the rancher to minimize the impact of cattle while allowing the ranching tradition to continue. I approached River Fork Ranch’s TNC manager, Lori Leonard, several years ago about the possibility of doing some acoustic monitoring at the ranch. She has been very welcoming of a variety of inventory and monitoring projects, and was very supportive of my idea.
Although the wetlands on the property are rich in wildlife, the location is surrounded by one major and a couple of minor highways (1-3 miles distant), a small but very active airport 4 miles away, and a commercial air route overhead. In other words, there is a lot of anthropogenic noise. At times, the noise from the highway or aircraft overhead nearly drowned out the sounds of the wildlife in the wetlands. In addition, the proximity to the mountains meant strong afternoon winds almost daily. So I wasn’t sure how well an acoustic monitoring program at this location would work.
I decided to focus my monitoring not so much on who was there, but when various species called throughout the spring and summer (acoustic phenology). Recording generates a lot of information (and huge files), so I decided to sample 10 minutes of every hour, starting an hour before sunset and running until a couple of hours after sunrise. This encompasses the most active periods for the species I was interested in (birds and frogs), and skips the daytime hours when human noise can be overwhelming. I set out recorders from April to November 2020, and March to September 2021. In 2020, I used a couple of different recorders, each was deployed for a night and collected the next day. The time-consuming driving and hiking to set up the recorders resulted in longer periods between recording than I wanted. So, in 2021, I deployed a recorder that was more programmable and had a very long battery life, meaning I only needed to check on it a couple of times during the summer. All of the recorders did a good job recording the species in the wetland.
Manually going through hours and hours of recordings to identify each species calling can be time consuming and tedious. But carefully reviewing the recordings identified at least 46 species of birds and 3 species of frogs. A number of companies have developed software now that allow automated species identification from recordings. One such is Arbimon (https://rfcx.org/ecoacoustics), a web-based program that conducts species identifications and soundscape analysis. The program identifies species by comparing a user-supplied template with the recordings. It then flags any snippet in the recordings that it thinks is a match. The researcher verifies the species as “present” or “not present” in each snippet. So, although automated, it still takes considerable time to identify suitable templates and assess all of the snippets.
As a proof of concept, I compared the results of manually identifying species in the recordings with the results from Arbimon for 6 species, for which I’ll present 4 here. For each 10 minute recording session, I recorded a “1” if the species called, and a “0” if it did not. I then summed the “1”s for each recording day. Because the number of 10 minute sampling periods varied throughout the season (based on sunrise/sunset times), I divided the sums of the 10 minute periods by the number of periods for each sampling day, which I refer to as an “activity index.” So if I detected a species during 3 sampling periods in a day, and there were 12 sampling periods for that date, the activity index was 0.25.
Sandhill Cranes (Antigone canadensis)
The birding guides show Sandhill Cranes as migratory through this area, but the acoustic data (and observations) revealed that they are here throughout the summer, and probably breed here. It may be that most migrate through and only a few stop. They are considered “state sensitive.” Although the rattling call of the cranes was very distinctive to me in the recordings, Arbimon struggled with it.
American Bittern (Botaurus lentiginosus)
American Bitterns are somewhat secretive, and appear to be in the Carson Valley only during the breeding season. They often called near the recorder, and Arbimon did well at identifying their unique “oonk-GA-loonk” call (listen for it in the recordings below).
Sierran Tree Frog (Pseudocris sierrae (formerly regilla))
Sierran Tree Frogs are common residents in wetlands and puddles throughout the area. Their “ribbit” call is a harbinger of spring. Arbimon struggled a little with the calls, sometimes confusing them with the “kiddik, kiddik” of the Virginia Rail. Both types of species identification found a strong seasonality in their calls, with most of the activity done by early July, and minor amounts extending into the fall.
Greater Yellowlegs (Tringa melanoleuca)
Greater Yellowlegs migrate through the Carson Valley in the spring and again in the fall, between their wintering areas in Mexico to southern South America and their breeding grounds in Canada and Alaska. Their calls were soft and quick, and Arbimon had a little trouble with them.
Conclusion
The recordings were remarkably successful at identifying the species present in the wetland. They also identified season patterns in calling for most species, as well as year-to-year changes for some. Further research is needed to determine if outside factors (weather, etc.) affected year-to-year differences in calling activity, or if it was a sampling artifact.
Several species observed in the area were either not or very seldom detected in the recordings (Brewer’s sparrow, White-faced Ibis, Cinnamon Teal), but they might have been just out of range of the recorder. For some species, such as Sandhill Cranes and Bald Eagles, Arbimon really struggled with identifying the calls. For these species, visual surveys may be better at detection. Arbimon also appeared to be hampered by the anthropogenic and wind noise on the recordings. This will be addressed in the future by searching for better templates in the recordings, and minimizing wind noise on the microphone.
The cacophony of the dawn chorus also impacted the ability to identify individual calls, both by myself and Arbimon. This can be heard in the following 2 recordings from May 8, 2020, one pre-dawn (about 4:30 am) and one at dawn (around 5:30 am):
The monitoring program is continuing for 2022 at the same location. I hope to have the recorder out in the field for a longer period (March to October), and conduct sampling more frequently to get a finer resolution and species coming and goings.
Canada Geese and goslings. One of many species that breed at River Fork Ranch, NV.
A wonderful post and contribution to the field. Thank you.
Thanks!