Aim: North America’s scoter species are poorly monitored relative to other waterfowl. Black (Melanitta americana), surf (M. perspicillata), and white-winged (M. deglandi) scoter abundance and trend estimates are thus uncertain in many parts of these species’ ranges. The most extensive source of waterfowl abundance and distribution data in North America is the Waterfowl Breeding Population and Habitat Survey (WBPHS). Although the WBPHS effectively monitors most species, both its timing and geographic coverage may not allow for accurate scoter monitoring. Therefore, our goal was to better define when and where scoters breed to help interpret survey results and optimize survey methods for scoters.

Location: Canadian boreal shield, taiga shield, and low-Arctic tundra; Alaska.

Taxon: Scoters (Genus: Melanitta)

Methods: We integrated satellite telemetry tracking data from scoters marked at multiple molting, staging, breeding, and wintering areas along the Atlantic and Pacific coasts to quantify continent-wide breeding chronology and distribution. We also examined possible drivers of variation in timing of arrival, length of stay, and departure at nesting locations.

Results: We documented a northwest-to-southeast distribution of breeding sites across Alaska and Canada. On average, scoters arrived at nest sites on June 1. Surf scoters arrived earliest, stayed for shorter periods, and departed earliest. Pacific-wintering scoters began breeding earlier than Atlantic-wintering birds. Additionally, birds arrived at nesting locations earlier in years with earlier snowmelt, and later snowmelt reduced lengths of stay for males. Breeding chronology also varied by age group, with adults arriving earlier than subadults.

Main conclusions: Our study is the first to comprehensively describe spatial variation in timing of breeding of both Atlantic and Pacific populations of all three scoter species across North America. Our results increase our understanding of how current surveys enumerate scoters and will inform possible supplemental efforts to improve continental monitoring of scoter populations.

Dataset 1. Scoter satellite telemetry breeding location and chronology data

We compiled data for a total of 94 black scoters, 124 surf scoters, and 95 white-winged scoters captured between 1999 and 2019 at sites in the Great Lakes and along the Atlantic and Pacific coasts. All captures were made during scoter molting, staging, and wintering periods (August - March). Birds were implanted with Platform Transmitter Terminals (PTTs), programmed to collect location information once every three to five days. Detection data were filtered to only include detections from 20 April to 1 July of each year (i.e., the suspected nesting period of North American scoters). We identified clusters of repeated detections where individuals settled for at least nine days within the potential breeding range. We identified nesting locations by drawing a 20 km buffer around the centroid of each cluster, and all detections within each 20 km buffer were considered movements within the nesting location. We then classified arrival date as an individual's first detection at the nesting location, departure date as the last detection at the nesting location, and length of stay as the number of days between an individual's last and first detection. 

Dataset 2. Spatial predictions of scoter nesting location arrivals

We used linear mixed-effects models to evaluate whether variations in scoter arrival dates at nesting locations differed by species, wintering origin, age class, sex, and/or snow-free date. Here, snow-free date was estimated at the nesting location of each PTT-marked individual using the National Oceanic and Atmospheric Administration (NOAA) National Snow & Ice Data Center’s Ice Mapping System (IMS) snow cover maps. We downloaded daily snow cover data for Canada and Alaska for all years with detection data (1999–2019) at a spatial resolution of 24 km. Snow-free date was identified for each 24 km square as the third consecutive day of the year (DOY) with no snow. We then calculated the snow-free date for each year at each nesting location as the mean DOY of snowmelt within a 24 km square buffer around each nesting location centroid.

As a whole, estimated scoter breeding locations closely followed the delineation of the Taiga ecoregion, with additional detections in the Hudson Plains ecoregion and along the northwest coast of Alaska. We predicted each species’  arrival timing across this area using the best-supported model for scoter arrival dates, based on the lowest Akaike’s information criterion corrected for small sample size (AICc), and annual snowmelt data. First, we created a polygon for prediction (hereafter the ‘prediction polygon’). We downloaded Level 1 ecoregion shapefiles from the U.S. Environmental Protection Agency. We drew a 100 km buffer around the Taiga and Hudson Plains ecoregions to account for scoters potentially nesting in ecoregion transition zones. We then extended this polygon to include PTT-estimated breeding locations along the northwest Alaskan coast. We used the same polygon for all three species. Next, we estimated arrival dates for each species across the prediction polygon. To do this, we overlaid the prediction polygon with a hexagonal grid. Snow cover data had a spatial resolution of 24 km; we thus generated grids where each hexagon had an apothem (i.e., distance from the hexagon centre to side) of 24 km. Using annual snowmelt rasters from 2010 to 2020, we generated a new raster showing mean snowmelt date for this period (roughly the last 10 years of this study), which was used to determine mean snow-free date within each hexagonal grid cell. We then used each hexagon’s estimated mean snow-free date in our best-supported model to estimate arrival dates in each hexagon. For prediction, we held wintering origin at Atlantic, age at adult, sex at male, and year and animal identity at 0 (i.e., reference levels). All spatial analyses were completed using the “sf” and “raster” packages.  

Dataset 3. Brood observations

We collected brood observation data to evaluate whether PTT implantation delayed PTT-marked birds. Brood observation data were collected from various sources. We collected Labrador brood data from unpublished reports (Barrow, 1982; LGL Limited, 2008) and surveys (SGG). Québec data came from various reports and publications (Benoit, 2005; Benoit et al., 1994, 1996; Consortium Gauthier & Guillemette - G.R.E.B.E., 1990; Morneau, 1998; Morrier et al., 2008) and unpublished survey data (J-FP and CL). Information on northern Ontario broods came from Brook et al. (2012). J. Fisher (USFWS) provided a large dataset of brood observations collected in Alaska from 1990–1993. We estimated nest initiation dates (i.e., date the first egg was laid) by back-dating from the date of brood observation, using the median age of each duckling plumage age class (from Gollop & Marshall, 1954; adjusted for surf scoters according to Lesage et al., 1997), a laying rate of one egg per day, and a mean clutch size of 7 eggs. In total, nest initiation dates were estimated for 452 black, surf, and white-wing scoter broods observed between 1989 and 2012.

Data files can be opened with Microsoft Excel.