Your browser doesn't support the features required by impress.js, so you are presented with a simplified version of this presentation.

For the best experience please use the latest Chrome, Safari or Firefox browser.

Core Questions & Objectives

  • Which species can be considered “boreal ducks” or “prairie ducks”?

  • How do we identify a boreal duck specialist?

  • Assess the importance of the boreal and prairie-parkland regions to each species, using models built from the Waterfowl Breeding Population and Habitat Survey

  • Sub-objective Highlight the importance of the boreal forest for waterfowl and encourage discussion and more research in the region



TS: We expect that species’ distributions/ranges and abundances are outcomes of selection for and adaptation to suitable environmental conditions.

  • Hutchinson (1957) defined A species fundamental niche represents the range of environmental conditions that it can survive and reproduce within.
  • The realized niche represents a reduced set of conditions in response to competition and other biotic interactions.
  • A species range reflects its fundamental and realized niches, while the distribution of its population within its range is assumed to reflect habitat selection / the habitat within which the species shows optimal survival and reproductive success. - maybe check Gaston..
  • (Note: no discussion of ecological traps. we are working under the assumption that density = better survival / reproduction = better habitat.)
  • Need to find some citations for these sections
  • This is true at large scales, which delineates the range of the species
  • And at smaller scales, which determines the distribution of populations within that range
  • “what determines the range of habitats in which a species occurs, and how does each individual determine when it’s in an appropriate habitat? The first question is evolutionary: how has natural selection shaped habitat choices? The second is behavioral: what cues does a bird use in “choosing” its home? We put choosing in quotes to emphasize the presumed absence of conscious choice. Indeed, some ecologists employ the term “habitat use rather than “habitat selection” to avoid the connotation of birds making deliberate decisions among habitat alternatives.” (
  • In the simplest case, species are at equilibrium with their environment, such that they occur, survive, and successfully reproduce offspring in regions with conditions they are well-adapted to, and not in regions with conditions they are poorly-adapted to (Southwood 1977, Cody 1985). - Look up definition of “equilbrium” in context of SDMs and ensure I’m using it correctly. Add a citation that defines “equlibrium” to this sentence
  • Current patterns in distribution of species across a landscape are assumed to represent accumulated successes in survival, reproduction, and “choices” in habitat selection and dispersal

  • Check with Gaston notes for some citations

Natural selection can act on populations when there are differences in survival and reproduction among different habitats. Over time, species’ become adapted to particular ranges in habitat conditions in that they successfully reproduce and persist as populations in one set of conditions and not so outside of those conditions. The geographical area containing those conditions is the “range” (this idea based off a Martin 1998 citation in Clark & Hutler 1999 about natural selection occurring on nests in different habitats)

  • Seriously, do some cross-referencing on definition of “Range” - Gaston probably

TS: Historically, waterfowl species have been classified as either “prairie” or “boreal” species, meaning these species ‘represent’ or ‘are typical’ of a given region/biome.

  • For example, Bethke & Nudds 1993 refer to (15) species that are representative of three regions - mixed-prairie, aspen parkland or boreal - based on their overall breeding distributions from Bellrose 1980.

TS: We expect that nesting habitat, food resources, growing season, climate, and other factors differ between different ecological regions (e.g. between prairies and boreal forest), and might therefore lead to differences in survival among difference species.

habitat & food resources
Due to topographical, hydrological, and climatic differences, wetlands generally differ from the prairie and parkland to the boreal region.

  • Bethke & Nudds 1993 suggested that a gradient in habitat heterogeneity and productivity resulted from gradients in wetland variability and basin morphology and that this gradient in habitat conditions led to variation in duck diversity. (paper includes many citations)

Prairie/grassland wetlands are generally highly productive, due to climatic variability and periodic droughts leading to high vegetation in shallower wetlands (Sheehan et al 1987, Adams 1988). - even compared to parkland wetlands (see Bethke & Nudds 1995)

  • Bethke & Nudds 1993 highlight that the prairies show highly variable wetlands due to high variability in precipitation, low run-off, high summer temperature, and high evapotranspiration.
  • The abundance and diversity of wetlands in the prairie-parkland are determined largely by precipitation in the preceding years (Adams 1988 - cited in Bethke & Nudds 1995)
  • Northern wetlands, in contrast, are generally less variable due to opposite environmental conditions.
  • Topographically, wetlands in the south are typically shallower or “saucer-shaped” while northern wetlands are typically deeper or “bowl-shaped” (Sheehan et al 1987, Adams 1988, cited in Bethke & Nudds 1993). This gradient may be an oversimplification, since boreal wetlands range from shallow peatlands to deeper water bodies.

growing season and climate
The growing season is shorter further in the north, but so is the day length. Climatic conditions are more moist, less variable, and cooler (citation? - also consult with maps of environmental variables to make sure I’m correct).


TS: Our objective was to identify which species are highly adapted to boreal, hemiboreal, or prairie resources. [species that are not highly adapted to one region over the other can be considered “generalists”]

  • We assume that species that are adapted to one biome/region over another will encompass more of that biome/region within their range, and will show a disproportionate amount of their population within that region/biome as well.
  • To this end, we used predictive models of waterfowl abundance to quantify the proportion of each species’ range and population within each region. Ducks with a regional/ecozone affinity will show higher proportions of their ranges and/or populations in one region above all others.
  • As an extension, we followed a use:availability approach to correct for the relative sizes of the different regions.
  • This ratio provides an additional index of the relative affinity of a species towards a given region, as demonstrated by a disproportionate amount of its population within a region compared to its distribution across the various regions.



TS: Our study area represents the Canadian portion of the WBPHS (Figure already mentioned).

  • With the exception that part of stratum 12 was excluded due to the extent of one of the environmental variables.

TS: We defined our regions following a combination of Brandt’s (2009) and the Bird Conservation Regions [?]

  • Our boreal was delineated according to the boreal within Brandt (2009), excluding boreal alpine. This classification was based largely on _[whatever Brandt used] and this boreal delineation is largely adopted by boreal initiatives (BAM, BEACONs, EC’s caribou? CBFA? Other?)
  • Prairie-parkland was defined according to the BCR “prairie pothole”.
  • These two shapefiles were not all inclusive nor mutually exclusive. Overlapped areas were classified as boreal while gaps were classified as prairie-parkland.


  • Hemiboreal was also classified according to Brandt (2009), excluding alpine hemiboreal. Much of the hemiboreal is included within the traditional “prairie-parkland” or “prairie pothole” regions, and is similar to the parkland of Bethke & Nudds 1995. Some of the “Boreal transition Zone” (citation) is also included within the hemiboreal.
  • Prairie was classified according to the North American Level I Ecoregions. It represents mixed grassland prairie.
  • The remaining area of the study are was classified as “other”, but included Arctic, Northeastern Forests, Western Forests, and _.

Waterfowl Abundance Data and Maps

TS: We used waterfowl count data from the Waterfowl Breeding Population and Habitat Survey (WBPHS; Figure X).

  • Details of survey design and methods are described in Smith 1995, Zimmerman et al 2012, and Barker et al in prep.
  • We included segment-level counts of total indicated pairs from 1995-2010, excluding 2007 due to inconsistent sampling in that year (Barker et al in prep). This subset represents a relatively static, recent, and represent sample of waterfowl counts.

Figure Map of survey strata and transects

Layout of the Waterfowl Breeding Population and Habitat Survey, showing strata boundaries and flight transects.

TS: We built predictive models of waterfowl abundance using boosted regression tree (BRT) analyses.

[Brief description of methods, but refer to my in-prep paper]

Figure Map of predicted abundances for one species with delineated range

[species] pair abundance, as predicted from boosted regression trees (citation: Nic’s paper) and delineated range.

Range Delineation

TS: Range was delineated as the concave/alpha hull encompassing all densities above a species-specific density threshold.

We identified each species’ density threshold as follows.

  • We extracted the cell-level predicted abundances from each cell in the prediction raster.

  • We ordered the abundances from highest to lowest and then calculated the cumulative sum abundance for each cell.

  • We converted this cumulative sum abundance into a proportion by dividing by the total predicted population in the full study area.

  • We plotted the proportional cumulative sum abundance as a function of abundance and then calculated the first and second derivative of this function.

  • We used the lowest density associated with a change point (identified as a minimum in the first derivative and a point where the second derivative equals zero).

Using each species’ density threshold, we separated presences from absences within the prediction raster, and then calculated the concave (i.e. alpha) hull of the resulting presences. This technique creates a polygon that encompasses the external points in a cluster of points (citation).

  • We selected an alpha value of XX. This fairly low value delineated a range that encompassed much of the predicted presences without including the absences between them.

Descriptive and Selection Metrics

TS: We calculated the amount of each species’ predicted population and delineated range contained within the boreal and prairie-parkland regions as a proportion of the total predicted population and predicted area within the study area.

  • We also calculated the ratio of proportion population to proportion range to calculate the relative affinity of a species for a region.
  • Comparison of this ratio to 1 was used as an index of regional affinity.
  • Since species with nearly the full amount of their population or range in a region will result in a ratio of 1, we also considered any species with > XX % of their range and/or population within a region as having a regional affinity.

Null Distribution and Randomization Test


Descriptive Metrics

Selection Metrics and Randomization Test


Figure Point (or bar) plot showing the proportion of each species’ range within boreal, prairie-parkland, and “other” regions.

The proportion of each species’ range contained within the boreal, prairie-parkland, and other regions.

Figure Point (or bar) plot showing the proportion of each species’ population within boreal & prairies

The proportion of each species’ range contained within the boreal, prairie-parkland, and other regions.

Figure Point plot with abline at 0,1 to show where boreal ratio falls for each species. Colour-code/label specialization regions (e.g. above a given proportion and above/below 1 by a certain amount). Colour-code points by dabbler/diver

Boreal affinity of waterfowl species, where values above 1 indicate a boreal affinity and those below 1 indicate avoidance of the boreal. The green shaded region indicates the species classified as boreal due to the proportion of their range and/or population within the boreal.

Figure Point plot with abline at 0,1 to show where prairie-parkland ratio falls for each species. Colour-code/label specialization regions (e.g. above a given proportion and above/below 1 by a certain amount). Colour-code points by dabbler/diver

Prairie-parkland affinity of waterfowl species, where values above 1 indicate a prairie-parkland affinity and those below 1 indicate avoidance of the prairie-parkland region. The yellow shaded region indicates the species classified as prairie-parkland due to the proportion of their range and/or population within the prairie-parkland.


Patterns of abundance/distribution vs. process of natural selection

We modelled and based our analyses off observed patterns in species distribution and abundance. However, since we used a relatively long-term dataset (15 years), we indirectly asked about persistence, which yields some information about the process driving range determination. Species may occur in some locations, at low abundance or only in one of many years. This pattern would be reflected as very low predicted abundance, which would exclude it from the predicted range in our analysis. Ergo, our ranges are likely to capture some pattern in long-term persistence of a population.

  • Cross reference with the abundance doesn’t equal “suitable” habitat literature.

Summary of Results

Consistency in comparison with habitat preferences

  • Which species are identified as boreal vs prairie-parkland?
  • Are there patterns when classifying by feeding guild?
  • Are the patterns when classifying by nesting guild?

Comparison with conventional wisdom

[Are our results consistent with conventional wisdom regarding species for which the boreal is important? Prairies?]

Bethke & Nudds 1993 classify the following species as represenative of mixed prairie, aspen parkland, or boreal forest:

  • mallard (dab)
  • gadwall (dab)
  • wigeon (dab)
  • green-winged teal (dab)
  • blue-winged teal (dab)
  • northern shoveler (dab)
  • northern pintail (dab)
  • redhead (div)
  • canvasback (div)
  • lesser scaup (div)
  • ring-necked duck (div)
  • common goldeneye (div)
  • bufflehead (div)
  • scoters (div)
  • ruddy duck (div)
  • They separate between the mixed prairie and the aspen parkland, while we grouped these two regions together
  • They stated that gadwall, blue-winged teal, and northern shoveler were largely absent or infrequent in northern regions

Duck abundance was most strongly associated with conserved soil moisture in the grassland region compared to the parkland (Bethke & Nudds 1995). Ephemeral and temporary wetlands are more common in the grassland region than the parkland (Adams 1988), and are important food resources in spring (Krapu 1974). - cited in Bethke & Nudds 1995.
—> Doesn’t necessarily say anything about species regionalizations. Just that abundance is more dependent on wetland conditions/precipitation in grassland compared to parkland.

Practical Implications