Mallards have been replacing Black Ducks in human-altered landscapes in Ontario, 1996-2019

The American Black Duck (hereafter Black Duck) has received much attention over the years as a popular game bird species that experienced historical declines on its wintering grounds. Declines may be due to changes in the quality and quantity of breeding habitat, competition with Mallards, or both. Although the Black Duck population has stabilized, spatial variation in regional population trends of Black Ducks on their breeding grounds can provide insight into the relative importance of these alternative hypotheses at provincial and local scales. We analyzed indicated breeding pairs count data collected in Ontario between 1996 and 2019 through the Eastern Waterfowl Survey to evaluate the correlation in breeding densities and trends of the five most abundant waterfowl species. We also tested whether habitat and anthropogenic disturbance variables could explain the distribution of those species. Overall, Mallard breeding pair densities increased across the survey area, while Black Duck breeding pair densities decreased in the southern and northern part of the survey area but remained stable in the central part. Densities of breeding pairs of Black Ducks and Mallards were negatively correlated across survey plots but the trends in breeding densities were not correlated. Densities of Mallard breeding pairs increased in human altered landscapes whereas the density of Black Duck breeding pairs was negatively correlated with agriculture. Our results suggest that Mallards did not competitively exclude Black Ducks on the breeding grounds, but instead, Black Ducks were replaced by Mallards following encroachment of agriculture into Black Duck breeding habitat. Habitat management plans focused on breeding Black Ducks should emphasize reduction in the expansion of anthropogenic perturbations and increases in habitat conservation efforts in the central area of the breeding range in Ontario.

Methods
Waterfowl data
The waterfowl data have been collected through the The Eastern Waterfowl Survey (EWS). A breeding waterfowl survey that has been conducted annually since 1990 by CWS in an area extending from northeastern and central Ontario east through central and southern Quebec, New Brunswick, Nova Scotia, Newfoundland, and southern Labrador. Each spring between late-April and early-June, observers counted and identified ducks and geese observed on every waterbody, watercourse, and wetland within a 25-km2 square survey plot using a helicopter.

All waterfowl observations were identified to species, sex, grouping and numbers (e.g., lone or paired birds, flocked, etc.) and georeferenced to a location or waterbody. These observations were used to infer the number of Inidcated Breeding Pairs (IBPs) according to the standardized method developed by the Black Duck Joint Venture for helicopter surveys in Eastern Canada. The full methodology for the survey and the IBPs calculatons are described in Helicopter-based waterfowl breeding pair survey in eastern Canada and related studies.

Habitat Variables
The wetland habitat availability and landscape disturbance were extracted from the land cover information from the Ontario Forest Resource Inventory (hereafter Ontario FRI; available: https://www.ontario.ca/page/land-information-ontario). The Ontario FRI use imagery acquired from aerial photographs, combined with field validation, to produce digital thematic maps of classified land cover within the boreal-hardwood transition and boreal regions of Ontario.It has a multi-spectral digital orthography resolution of 40 cm and is updated every 10 years, with the most recent update in relation to our study period completed in 2011. We used ArcGIS (ESRI 2014. ArcGIS Desktop: Release 10.2.2. Redlands, CA: Environmental Systems Research Institute) to calculate the area and perimeter of features of biological interest within a 1 km buffer around each plot to account for landscape conditions.

Data
The explanatory variables have been scaled for the analysis. The raw data and the scaled data are included. 

Code
The R code and Stan code needed to run the model are also included.