Radar revelations: Insect availability influences parental provisioning in breeding Tree Swallows (Tachycineta bicolor)
Data files
Jan 27, 2025 version files 4.48 MB
-
datafordryad.csv
4.48 MB
-
README.md
5.34 KB
Abstract
Airspace habitat is essential foraging space for Tree Swallows (Tachycineta bicolor), which rely on flying insects as their main source of food. Insect availability can change quickly from hour-to-hour or day-to-day, however it is unclear whether insectivores primarily respond to changing atmospheric dynamics, resource dynamics, or the combination. Rapidly changing conditions are common in high elevation areas — an understudied portion of the Tree Swallow’s breeding range. To explore the relationship between food availability and high elevation weather conditions as related to female provisioning, we deployed a mobile radar unit to collect insect abundance data during the 2022 and 2023 summer breeding seasons at a high elevation site in Colorado’s Rocky Mountains. We monitored 41 active nest boxes using radio-frequency identification (RFID) technology to track female provisioning behavior. We deployed three models to assess (1) how strongly swallow provisioning rates correlated with insect traffic rates, (2) how well swallow provisioning rates were explained by insect traffic rates and weather conditions, and (3) how insect traffic rates were related to weather conditions. Although there remains substantial unexplained variation in tree swallow provisioning rates, we found a significant positive relationship with insect traffic rate, negative relationship with precipitation, and curvilinear relationships with temperature and wind speed. Weather variables and time of day explained nearly 80% of variation in insect traffic rate, and the strength of these relationships suggest weather conditions serve as a good proxy of airborne insect activity. This research presents a link between our vast airspace habitat and animal ecology, advancing our understanding of how flying organisms respond to rapidly changing conditions in aerial environments and how multiple factors contribute to variation in provisioning rates in an aerial insectivore.
README: Radar revelations: Insect availability influences parental provisioning in breeding Tree Swallows (Tachycineta bicolor)
https://doi.org/10.5061/dryad.cnp5hqcfx
Description of the data and file structure
The goal of this project was to explore the relationship between food availability and high elevation weather conditions as related to female Tree Swallow (Tachycineta bicolor) provisioning behavior. By combining data on nest visitations, weather conditions, and insect activity, we sought to gain understanding of how airspace habitats are linked to behavioral ecology of aerial insectivores.
Files and variables
File: data_final.csv
Description: This file includes insect data collected by a BirdScan MR-1 radar, processed weather data, and Tree Swallow nest box visitation data.
Variables
- avg.temp: the average hourly temperature in degrees Celsius
- sum.precip: the total hourly precipitation in millimeters
- avg.wspeed: the average hourly wind speed in meters per second
- bin.precip: hourly precipitation as a binary factor (whether or not rain fell)
- date: the date that the measurement was observed
- month: the month that the measurement was observed
- day: the day that the measurement was observed
- DOY: the day of the year that the measurement was observed
- year: the year that the measurement was observed
- time: the time at which the measurement was observed
- sunrise: the time at which local sunrise occurred
- sunset: the time at which local sunset occurred
- percentile: the number of hours after local sunrise divided by the total number of daylight hours
- count: the hourly total number of visits a female Tree Swallow made to her nest box
- band: the USGS aluminum bird band identification number
- sqrt_insects_25_75_50: the square root of insect traffic rate (insects/km/hr) between 25 and 75 meters above ground
- sqrt_insects_75_125_50: the square root of insect traffic rate (insects/km/hr) between 75 and 125 meters above ground
- sqrt_insects_125_175_50: the square root of insect traffic rate (insects/km/hr) between 125 and 175 meters above ground
- sqrt_insects_175_225_50: the square root of insect traffic rate (insects/km/hr) between 175 and 225 meters above ground
- sqrt_insects_225_275_50: the square root of insect traffic rate (insects/km/hr) between 225 and 275 meters above ground
- sqrt_insects_275_325_50: the square root of insect traffic rate (insects/km/hr) between 275 and 325 meters above ground
- sqrt_insects_325_375_50: the square root of insect traffic rate (insects/km/hr) between 325 and 375 meters above ground
- sqrt_insects_375_425_50: the square root of insect traffic rate (insects/km/hr) between 375 and 425 meters above ground
- sqrt_insects_425_475_50: the square root of insect traffic rate (insects/km/hr) between 425 and 475 meters above ground
- sqrt_insects_475_525_50: the square root of insect traffic rate (insects/km/hr) between 475 and 525 meters above ground
- sqrt_insects_25_125_100: the square root of insect traffic rate (insects/km/hr) between 25 and 125 meters above ground
- sqrt_insects_125_225_100: the square root of insect traffic rate (insects/k/hr) between 125 and 225 meters above ground
- sqrt_insects_225_325_100: the square root of insect traffic rate (insects/km/hr) between 225 and 325 meters above ground
- sqrt_insects_325_425_100: the square root of insect traffic rate (insects/km/hr) between 325 and 425 meters above ground
- sqrt_insects_425_525_100: the square root of insect traffic rate (insects/km/hr) between 425 and 525 meters above ground
- sqrt_insects_25_225_200: the square root of insect traffic rate (insects/km/hr) between 25 and 225 meters above ground
- sqrt_insects_225_425_200: the square root of insect traffic rate (insects/km/hr) between 225 and 425 meters above ground
- sqrt_insects_25_325_300: the square root of insect traffic rate (insects/km/hr) between 25 and 325 meters above ground
- sqrt_insects_25_425_400: the square root of insect traffic rate (insects/km/hr) between 25 and 425 meters above ground
- sqrt_insects_25_525_500: the square root of insect traffic rate (insects/km/hr) between 25 and 525 meters above ground
- sqrt_insect: the square root of insect traffic rate (insects/km/hr)
- insect: insect traffic rate, or the number of insects passing through the radar beam in one horizontal kilometer per hour (insects/km/hr)
File: SupplementalTables_JAB.docx
Description: This file includes two supplemental tables. One shows AIC values for models predicting female provisioning rates as a function of insect traffic rates in different altitude bands. The other gives Pearson's correlation coefficients for each pair of altitude bands tested.
Code/software
File: JAB_script.Rmd
Description: This file includes the code used to build and run models of female provisioning rates as a function of insect traffic rates and weather conditions. It also contains code for models of insect traffic rate as a function of weather conditions as well as code for creating figures.
We used Program R (v4.1.2) to view and analyze these data. We loaded the following packages for analysis:
- birdscanR (Haest, et al. 2023)
- glmmTMB (Brooks, et al. 2017)
- splines
- lmtest (Zeileis and Hothorn 2002)
Methods
We monitored a system of 166 nest boxes at Colorado State University’s Mountain Campus (40.5705° N, 105.5913° W, 2743m elevation) weekly from early May through early August in 2022 and 2023 to collect data on female Tree Swallow provisioning behavior. We checked for nest building activity, and we increased visits to twice each week at active boxes once a complete nest was present. We tracked egg laying, and once a full clutch was present, we captured females inside the nest box as they incubated their eggs. We then banded them with a USGS aluminum federal band and attached a PIT tag (Eccel 2.3mm EM4102) to their other leg.
We deployed a BirdScan MR1 (Swiss Birdradar Solution AG) onsite within a 1-km distance from all nest boxes. The BirdScan MR1 operated in short-pulse mode, and we configured its settings to follow methods used in other BirdScan MR1 studies. We used reflectivity detections to calculate an hourly traffic rate, i.e., the number of insects passing through a 1-km horizontal diameter of detection around the BirdScan every hour. Traffic rates were calculated in 50m vertical intervals from 50-500 m above ground level.
We deployed RFID data loggers (Scissortail Electronics LLC, Noble, Oklahoma) on active nest boxes, which recorded PIT-tag IDs along with a timestamp whenever a tag was present at the nest entrance. The RFID readers were programmed using an Arduino-based platform. Antennas searched for a signal for 100 milliseconds and cycled into a sleep mode for 200 milliseconds to conserve power. It was common for swallows to perch on the nest box entrance for several seconds before entering. Although the readers were programmed to wait 8 seconds between recording consecutive tag reads, we still had to condense strings of continuous tag reads to prevent overestimating the number of female nest box visits. For this reason, we filtered out successive readings that occurred within a 2-minute window. For each female swallow, we obtained hourly visit counts on days when the swallow’s RFID tag was active and insect activity estimates were available from the BirdScan. For hours when the female was not detected entering or exiting the nest box, we recorded the visit count as zero.
We collected weather data using instrumentation on site, located approximately 1.5km from the farthest nest boxes and 0.75km from the BirdScan. This instrumentation collected data every five minutes on temperature (Campbell Scientific EE181 sensor), precipitation (Hydrological Services TB4/0.01 sensor, replaced with a Texas Electronic TE525WS rain gage on July 11, 2023), and wind speed (RM Young 5108 sensor). We aggregated these five-minute readings into hourly measurements, calculating the average hourly temperature (°C) and wind speed (m/s). We summed precipitation values and converted measurements into a binary factor (raining or not). To calculate scaled time of day relative to day length, we divided the number of hours after local sunrise by the total number of hours of daylight. Based on this calculation, proportions ranged from 0 to 1 with 0 corresponding to sunrise and 1 corresponding to sunset.