2  Sampling frame

Area of study

The study area for Quebec is outlined in Figure 2.1. It was expanded beyond the Boreal boundary to include the Arctic ecosystems. The study area contains a total of 7 ecozones, and their sizes and proportions are described in Table 2.1:

Table 2.1: Area (in hectares) and proportion of ecozones covered by the study area.

Ecozone	Area (% prop)
Taiga Shield	56275939 (35.74)
Boreal Shield	51342357 (32.61)
Southern Arctic	27378100 (17.39)
Northern Arctic	12918116 (8.2)
Hudson Plain	6235409 (3.96)
Arctic Cordillera	1713974 (1.09)
Atlantic Maritime	1587877 (1.01)

Figure 2.1: Study area (colored polygons) and its ecozone. The ecozone map was extracted from the Terrestrial Ecoregions of Canada data product - Government of Canada; Agriculture and Agri-Food Canada.

In order to accommodate habitat heterogeneity, the study area was hierarchically stratified into different levels of spatial aggregation. Below, we will provide a brief description of each of these strata, ranging from the ecoregion level to the specific sampling point level. For a more comprehensive explanation of the reasoning behind each stratification, please refer to Van Wilgenburg (2020).

Ecoregion

The ecoregion is the first level of aggregation in the sampling design. The sample size and habitat inclusion probability (described in the next chapter) are defined for each separate ecoregion. There are a total of 26 ecoregions in the study area (Figure 2.2), and their details are described in Table 2.2. Ecoregion 131 was excluded from the study area because it was too small to support enough sampling points for the random sampling design.

Table 2.2: Area (in hectares) and proportion of ecozones covered by the study area.

Code	Name	Area (% prop)
101	Central Laurentians	19431698 (12.35)
47	Central Ungava Peninsula	18114106 (11.51)
74	New Quebec Central Plateau	17262947 (10.97)
72	La Grande Hills	12929820 (8.22)
75	Ungava Bay Basin	9671186 (6.15)
103	Mecatina Plateau	9388108 (5.97)
100	Riviere Rupert Plateau	9083301 (5.77)
73	Southern Ungava Peninsula	8247511 (5.24)
96	Abitibi Plains	6787969 (4.31)
99	Southern Laurentians	5853410 (3.72)
48	Ottawa Islands	5766770 (3.67)
31	Northern Ungava Peninsula	5714604 (3.63)
28	Meta Incognita Peninsula	5295229 (3.37)
217	James Bay Lowlands	4482672 (2.85)
86	Mecatina River	2552186 (1.62)
46	Southampton Island Plain	2550507 (1.62)
76	George Plateau	2501318 (1.59)
30	Wager Bay Plateau	1908283 (1.21)
77	Kingarutuk-Fraser River	1861164 (1.18)
216	Hudson Bay Lowland	1752738 (1.11)
7	Torngat Mountains	1713974 (1.09)
117	Appalachians	1553719 (0.99)
78	Smallwood Reservoir-Michikamau	1148044 (0.73)
49	Belcher Islands	946717 (0.6)
102	Anticosti Island	790375 (0.5)
131	Iles-de-la-Madeleine	34149 (0.02)

The sample size for this study was determined solely based on the size of the ecoregion. While the BOSS design considered bird species richness to increase sampling in regions with more bird species, we chose not to use this metric because it may be biased by sampling efforts in the southern region, potentially increasing sampling bias in already well-covered regions. Our goal was to sample 2% of the available hexagons (PSU described below) in each ecoregion. We defined a hexagon as available for sampling if at least 20% of it contained natural habitat types.

Figure 2.2: Ecoregions (code) of the study area. Code description is detailed in Table 2.2.

Primary Sampling Unit (PSU)

We followed the BOSS design by using a 5 km diameter hexagon (Figure 2.3) as the Primary Sampling Unit (PSU). This was the lowest level of aggregation before performing the stratified sample with the GRTS algorithm. We selected only the hexagons whose centroid fell within the study area. Similarly, each hexagon was classified into one of the ecoregions using the same centroid rule. The number of hexagons, available hexagones, and sample size is described in Table 2.3.

Table 2.3: Distribution of total hexagons, hexagons with at least 20% of natural habitats, and sample size (2%) across the ecoregions.

Ecoregion code	Ecoregion name	Total Hexagons	Available Hexagons	Sample size
7	Torngat Mountains	866	866	17
28	Meta Incognita Peninsula	65	65	1
30	Wager Bay Plateau	174	174	3
31	Northern Ungava Peninsula	2307	2307	46
46	Southampton Island Plain	206	206	4
47	Central Ungava Peninsula	9580	9580	192
48	Ottawa Islands	36	36	1
49	Belcher Islands	4	4	0
72	La Grande Hills	7490	7490	150
73	Southern Ungava Peninsula	5019	5019	100
74	New Quebec Central Plateau	10773	10773	215
75	Ungava Bay Basin	5549	5549	111
76	George Plateau	1482	1482	30
77	Kingarutuk-Fraser River	1184	1184	24
78	Smallwood Reservoir-Michikamau	715	715	14
86	Mecatina River	1607	1607	32
96	Abitibi Plains	4139	4139	83
99	Southern Laurentians	3563	3563	71
100	Riviere Rupert Plateau	5539	5539	111
101	Central Laurentians	11970	11970	239
102	Anticosti Island	487	487	10
103	Mecatina Plateau	5711	5711	114
117	Appalachians	957	957	19
131	Iles-de-la-Madeleine	10	10	0
216	Hudson Bay Lowland	23	23	0
217	James Bay Lowlands	2293	2293	46

Secondary Sampling Unit (SSU)

For each PSU hexagon, we created a grid of Secondary Sampling Units (SSUs). The SSU represents the ultimate sampling locations to be utilized in the field. Instead of using the proposed 300-meter distance from the BOSS design, we followed the approach used in the Ontario regionalization design, where each SSU was separated by 294 meters. We made this small reduction in distance to ensure that the same number of SSUs were present across all PSU hexagons. Figure 2.3 displays the distribution of Secondary Sampling Units (SSUs) within a hexagon.

Figure 2.3: Distribution of Secondary Sampling Units (SSUs) spaced 294 meters apart within an Primary Sampling Unit (PSU) hexagon.

Van Wilgenburg, C. Lisa AND Campbell, Steven L. AND Mahon. 2020. “A Cost Efficient Spatially Balanced Hierarchical Sampling Design for Monitoring Boreal Birds Incorporating Access Costs and Habitat Stratification.” PLOS ONE 15 (6): 1–28. https://doi.org/10.1371/journal.pone.0234494.