Data from: Estimating waterfowl breeding pair and brood densities using distance sampling with uncrewed aerial systems
Data files
Dec 12, 2025 version files 1.19 MB
-
BroodData.csv
913.72 KB
-
PairData.csv
275.24 KB
-
README.md
2.52 KB
Abstract
Wildlife management relies on effective methods to estimate population indices, which are crucial for understanding the status and trends of wildlife populations and the factors that influence their persistence. Biologists have conducted surveys of waterfowl since the 1950’s using a combination of aerial (fixed wing aircraft) and ground-based methods to account for imperfect detection. While some individuals may be missed by aerial observers, ground observers are assumed to detect all individuals. This allows for a visual correction factor to be applied to the initial aerial count. However, the assumption that ground-observers are detecting every individual is unlikely and can potentially lead to biased population estimates.
We conducted breeding waterfowl surveys in Wisconsin during 2022 and 2023 and recorded observations of duck pairs and broods using uncrewed aerial systems (UAS) equipped with thermal cameras, which have been shown to improve detection of many wildlife species. We flew UAS surveys along transects and employed distance sampling to estimate duck densities while accounting for imperfect detection. This can be accomplished by recording the distance at which birds are detected and fitting a detection function to the data, which models detection probability and excludes the need for two independent observers. Our overall goal was to determine if UAS distance sampling can be used to monitor breeding waterfowl and produce reliable population metrics.
The archived datasets contain transect-based observations of waterfowl pairs and broods collected over both years of the study, along with their associated distances from the observation transect. Additional observational variables are also included to support analyses in this study.
Dataset DOI: 10.5061/dryad.wh70rxx2m
Description of the data and file structure
The pair and brood data files contain observational data collected during both years (2022 & 2023) of UAS surveys in Wisconsin. Each row in the file represents a single event and includes associated survey information. If multiple observations were made on a transect during a single survey, there will be one row per observation. A positive detection is indicated by a "1" in the "Pairs" or "Broods" column, while transects with no observations have a single row with a "0" in the "Pairs" or "Broods" column.
Files and variables
File: BroodData.csv
Description: UAS brood observations during the 2022 and 2023 field seasons
Variables
- Effort: length of transect (km)
- Region.Label: survey ID
- Sample.Label: transect ID
- object: unique identifier for each row of data
- Study.Area: name of the study area
- Area: total area surveyed (square km)
- distance: perpendicular distance between the observation and the transect it was observed from
- Date: date of the survey (m/d/y)
- Broods: brood observation (1 = yes, 0 = no)
- BroodAge: age class of observed brood (1 = downy young, 2 = incompletely feathered young, 3 = completely feathered young prior to flight)
- Behavior: observed behavior upon detection (F = flight, L = look up, N = no change, NA = no behavior recorded, S = swim away)
- Species: species of waterfowl observed
File: PairData.csv
Description: UAS pair observations during the 2022 and 2023 field seasons
Variables
- Effort: length of transect (km)
- Region.Label: survey ID
- Sample.Label: transect ID
- object: unique identifier for each row of data
- Study.Area: name of the study area
- Area: total area surveyed (square km)
- distance: perpendicular distance between the observation and the transect it was observed from
- Date: date of the survey (m/d/y)
- Pairs: breeding pair observation (1 = yes, 0 = no)
- Behavior: observed behavior upon detection (F = flight, L = look up, N = no change, NA = no behavior recorded, S = swim away)
- Species: species of waterfowl observed
Code/software
The datasets can be viewed in Microsoft Excel or read into Program R for analysis.
Data analysis was conducted in Program R version 4.3.1 using the ‘Distance’ package (Miller et al., 2019).