The following is the established format for referencing this article:Carpenter, T. M., C. L. Mahon, E. Bayne, J. L. Keim, and S. E. Nielsen. 2022. Influence of in situ oilsands development on occurrence of an avian peatland generalist and specialist. Avian Conservation and Ecology 17(2):23.
ABSTRACTDemand for petroleum products is causing habitat loss, alteration, and fragmentation of boreal forests in western Canada. Associated exploratory and extraction activities from in situ oil sands leave a network of (1) permanent polygonal features (e.g., processing facilities, extraction sites, gravel pits); (2) permanent linear features (e.g., roads, pipelines); (3) temporary polygonal features (e.g., exploratory well sites); and (4) temporary linear features, (e.g., winter roads, seismic lines). We examined the impact of these different types of disturbances on the occurrence of a generalist, the Dark-eyed Junco (Junco hyemalis) and a peatland specialist, the Palm Warbler (Setophaga palmarum) for an in situ development-lease area in northeast Alberta, Canada. Our goal was, first, to determine if energy development features have positive, negative, or neutral effects on species occurrence, beyond the amount of habitat disturbed, and second, to identify the relative impact of different types of energy features. Permanent polygonal features negatively influenced both species, decreasing the odds of occurrence for Dark-eyed Juncos by 0.63 times and 0.44 times for Palm Warblers for every 10% increase in percent area. However, permanent linear features increased odds of occurrence for Dark-eyed Juncos by 1.73 times and Palm Warblers by 3.93 times. We speculate that permanent linear features increase vegetation heterogeneity or influence insect-prey availability through edge effects. Although permanent polygonal and linear features had opposite effects (negative and positive, respectively), they had a similar relative effect on occurrence for both species. There were no effects of temporary polygonal or linear features on occurrence for either species examined here. Whereas our understanding of birds in boreal peatland forests is limited, these results are consistent with studies that suggest permanent linear features have more substantive local scale impacts than temporary disturbances.
Understanding species’ responses to anthropogenically derived habitat loss and fragmentation is a central issue to biodiversity conservation and management. As human populations grow, so does the need for natural resources with a concomitant increase in the intensity of anthropogenic footprint (Northrup and Wittemyer 2013). In the boreal forest of northern Alberta, demand for non-renewable energy products is contributing to alteration and perforation of forest habitats (Jordaan et al. 2009, Pickell et al. 2014, Rosa et al. 2017, Riva and Nielsen 2020). Alberta produced 3.1 million barrels of crude oil per day in 2019 (a 1.8% increase from 2018), representing 89% of Alberta’s $5.9 billion in non-renewable energy revenue (Alberta Government 2020). Crude oil in Alberta comes primarily from the extraction and processing of bitumen. Historically, bitumen in the Alberta oil-sands area was extracted through surface mining, i.e., digging for reserves within 75 m of the earth’s surface. However, less than 3% of bitumen deposits within the oil-sands region are surface-mineable (Alberta Government 2017). In situ, or below-ground, extraction methods have facilitated efficient access to the remaining 97% of the 142,200 km² extent of Alberta’s oil-sands area (Schneider and Dyer 2006, Alberta Government 2017). The expanding use of the steam-assisted gravity drainage (SAGD) extraction method (i.e., using below-ground injection of high-temperature steam to enhance the flow of bitumen for removal) requires locally intensive infrastructure for the extraction and processing of bitumen, such as central processing and steam-generating facilities, gravel-excavation pits, extraction-well sites, pipelines, powerlines, and road networks, hereafter called permanent features.
Prior to extraction and processing, extensive exploratory geological surveys are required to determine the depth and extent of underground bitumen reserves. Exploratory surveys involve clearing of forest vegetation to enable access of survey machinery. This habitat alteration leaves a footprint of different sizes and shapes with varying levels of remaining vegetation. These include temporary polygonal exploration-well sites, approximately 50 × 80 m vegetation clearings, linear winter roads or conventional seismic lines that are 5 to 20 m wide, and low-impact (three-dimensional or 3D) seismic lines that are 2 to 3 m wide and spaced every 30 to 100 m in a dense grid (Fig. 1). Exploratory features are considered by many to be temporary and low-impact relative to permanent features. Temporary features are typically left to regenerate naturally, although there is an increasing effort to plant trees on abandoned seismic lines. However, there is evidence that such features will persist, unregenerated, for extended periods (MacFarlane 2003, Lee and Boutin 2006, Van Rensen et al. 2015). Because of their high density, seismic lines have a greater impact on landscape-structure metrics, including the number of habitat patches, mean patch size, and amount of edge, than other linear features (Pattison et al. 2016). Projected increases in the proportion of extraction using in situ methods could increase the density of seismic lines from approximately 1.5 km per km² for conventional extraction methods to approximately 8 km per km² within the next 100 years, if all of Alberta’s bitumen deposits were developed (Schneider et al. 2003). Furthermore, localized areas requiring high intensity 3D seismic exploration can reach densities up to 40 km per km² (Stern et al. 2018).
Permanent and temporary features cause habitat loss and fragmentation for many wildlife species, influencing processes such as home-range selection (Tigner et al. 2015), movement patterns (Tigner et al. 2014, Dickie et al. 2017, Riva et al. 2018a), reproductive success (Ludlow et al. 2015, Bernath-Plaisted and Koper 2016), predator avoidance (Mumma et al. 2017), and foraging behavior (Scrafford et al. 2017). Even temporary features may be perceived by and alter the behavior of wildlife for extended periods of time (Tigner et al. 2014). For birds, the influence of energy-development features on abundance, density, and nesting success is species- and feature-specific (Thomas et al. 2014, Ludlow et al. 2015, Bayne et al. 2016, Farwell et al. 2016). Previous work on the effects of linear features on birds has been primarily in upland forests, where temporary development features may be quickly recolonized by deciduous vegetation. Avian responses in lowland black spruce (peatland) habitats, where vegetation succession occurs more slowly, are less well studied (Lee and Boutin 2006, Morissette et al. 2013, Van Rensen 2014). Unique peatland characteristics including high moisture organic matter, poor nutrient regimes, and moss-dominated ground cover, contribute to complex successional pathways following disturbance in these systems. At the same time, density of trees in natural peatlands is often considerably lower than in uplands. Thus, the narrow gaps created by linear features may not be perceived by birds as fragmenting these types of habitats. Therefore, understanding the relative impacts of different feature types on individual species in peatland habitats is important to enhancing our knowledge of the impacts to birds, particularly in habitats where the recovery of vegetation after development may be constrained by energy-sector activities.
Permanent and temporary development-feature types may have different effects on the occurrence of a species because of both the characteristics of the feature itself, and the influence of that feature on the surrounding habitat. Changes in vegetation structure can alter the microclimatic (e.g., temperature, moisture), resource availability (e.g., insect abundance), or biological (e.g., nest predation rates) influences associated with edge effects (Ewers and Didham 2006, Fischer and Lindenmayer 2007, Prevedello and Vieira 2010). The structural contrast of an anthropogenic feature may be a good predictor of the direction and magnitude of the expected edge effects (Prevedello and Vieira 2010). For forest ecosystems, edge effects are strongest when the structural contrast of the development feature is high relative to the unaltered habitat (Kennedy et al. 2010). Ovenbirds (Seirus aurocapilla), for example, showed an increased probability of including linear features within territories for lines with smaller widths (Bayne et al. 2005a) and greater vegetation cover (Lankau et al. 2013). Strength of an edge response, therefore, may be mediated by the degree of change in the structure of different development feature types. Larger, more permanent development features not only lead to greater area of resource change, but may also have greater contrast, enhancing differences in microhabitats such as sun exposure, temperature, or wind (Delgado et al. 2007, Stern et al. 2018). Thus, stronger edge effects may lead to a greater impact on the quality of the surrounding habitat.
Specialist species, i.e., those requiring a narrow set of resource attributes, may be more sensitive to habitat disturbance than generalist species, i.e., species with a broad ecological niche that utilize a wider range of resources (Clavel et al. 2011, Carrara et al. 2015). For example, in mixed hardwood- and oak-dominated (Quercus species) forests, specialized forest-interior species were less abundant at well sites than reference locations and showed declining abundance with increasing well-site density at the 25-ha scale, whereas generalist early successional species were more abundant at well sites than reference sites (Thomas et al. 2014). Furthermore, specialist species may be more sensitive to the characteristics of the altered habitat. Movement of forest specialists, for example, may be impeded by contrasting development features, potentially restricting specialists to larger patches of intact forest (Gillies and St. Clair 2010, Smith et al. 2011, Betts et al. 2014).
We investigate whether a conifer generalist and specialist respond differently to in situ oil-sands disturbance features of different sizes and types in peatland habitats. In northern Alberta, Dark-eyed Juncos (Junco hyemalis) and Palm Warblers (Setophaga palmarum) are both abundant in lowland habitats (Mahon et al. 2016, ABMI 2017). Dark-eyed Juncos are common to both disturbed and undisturbed coniferous habitats across the boreal forest region, and are considered to be conifer generalists (Schieck and Song 2006, Handel et al. 2009, Mahon et al. 2016). Palm Warblers are closely associated with lowland black spruce habitats, including bogs and fens, are identified as a specialist species in the boreal forest (Calmé et al. 2002, Mahon et al. 2016), and are thought to be sensitive to disturbances in these systems, requiring large tracts of intact peatland habitat (Calmé and Desrochers 2000, Poulin et al. 2006). Although both species are ground nesters common in treed-bog habitats, differences in the breadth of their habitat-use strategies provide an ideal comparison for contrasting avian responses to different anthropogenic features created by multi-feature SAGD disturbances.
We had two main objectives in this study: (1) determine if in situ, oil-sands development features contribute to positive, negative, or neutral effects on occurrence, beyond the influence of amount of habitat alone; and (2) determine the relative effect of different types of energy features on Dark-eyed Junco and Palm Warbler occurrence. We hypothesized that a conifer-generalist junco would show greater tolerance to development features than the peatland-specialist warbler because of the potential for complementary use of resources in the anthropogenically derived habitat, i.e., a greater ability for juncos to use novel resources provided by the development feature. Specifically, we hypothesized that Dark-eyed Juncos and Palm Warblers respond differently to permanent and temporary features because of (1) vegetation structural differences of the feature types, where permanent (non-vegetated) and temporary (containing early seral regenerating habitat) features provide different habitat; (2) feature size-related effects, where smaller features, such as seismic lines or other linear features, may not be perceived differently from natural vegetation gaps, whereas larger features would be more distinct and recognized; and (3) microhabitat conditions, where the unique differences in habitat conditions of each feature create favorable or unfavorable conditions. We anticipated that any or all of these responses could be present, and by identifying the strength and direction of responses to the features included in top models we would gain a better understanding of the relative impact of different feature types to these species.
This study was situated in a 17,000-ha in situ lease area, located approximately 30 km northwest of Fort McMurray, Alberta (Fig. 2). At the time of the study, the lease was in the early stages of development for Steam Assisted Gravity Drainage (SAGD) bitumen extraction. As a restricted access area, activity within the lease was specific to SAGD development and therefore provided a unique opportunity to examine SAGD development effects in isolation from other resource-sector activities. Upon completion of development, anthropogenic features in the lease will include approximately 3520 ha of exploratory seismic lines, temporary and permanent production well sites, winter roads, permanent gravel roads, pipelines, and other industrial facilities. During the study, all features were less than eight years of age. Permanent development features were built on a gravel surface (not vegetated), whereas temporary disturbances and seismic lines were vegetated and had variable plant composition and structure pending specific site and disturbance conditions. Within lowland habitats, these features were in the early stages of regeneration, exhibiting little to no regrowth of vegetation beyond the forb-herb and low shrub stage.
Occurring within the boreal mixedwood ecological region (Beckingham and Archibald 1996), the habitat within the in situ lease area primarily consists of lowlands, including open, shrubby, or treed bogs and fens. Vegetation communities in the nutrient-poor, acidic conditions of bogs are composed mainly of black spruce (Picea mariana) in the canopy and shrub layers. Dominant groundcover species include Labrador tea (Ledum groenlandicum), leatherleaf (Chamaedaphne calyculata), and bog cranberry (Vaccinium vitis-idaea). Nutrient-rich fens are more diverse and are composed of black spruce, tamarack (Larix laricina), alder (Alnus), birch (Betula), and willow (Salix) species in the main foliage layers. Low ground cover in fens may include sedges (Carex) and horsetails (Equisetum). Other less common habitats in the study area include mesic deciduous or coniferous-dominated upland forests, with leading canopy species including trembling aspen (Populus tremuloides), balsam poplar (Populus balsaminifera), or white spruce (Picea glauca). Low-bush cranberry (Viburnum edule) and prickly rose (Rosa acicularis) are common mesic shrubs. Drier or low-nutrient upland sites are dominated by jack pine (Pinus banksiana) or black spruce canopies and may include sparse blueberry (Vaccinium myrtilloides), Canada buffaloberry (Shepherdia canadensis), or Labrador tea in the understory.
Standard avian point-count techniques (Ralph et al. 1995, Matsuoka et al. 2014, Turgeon et al. 2017) were used to sample Dark-eyed Junco and Palm Warbler occurrence. Potential point-count sampling sites were identified following a stratified random selection procedure. Potential sampling stations were placed randomly across the study area between 0 and 6 km from permanent development features. The locations were spaced a minimum of 300 m apart (greater than the average detection radius for most songbird species) in attempt to maintain independence between sites (Matsuoka et al. 2014). Sampled survey sites were selected from the potential stations with a stratified random-selection procedure using imagery of current development and observer judgement in the field. Observers attempted to select survey routes that achieved a balanced sample size of point-counts across different development-feature types, total development-feature intensity, and given distance from permanent development features, which represent the greatest intensity of SAGD development features within the study area. Sample sizes achieved for total development intensity (% area) within 100 m included 66 sites with 0% development, 29 with > 0–10%, 18 with > 10–20%, 96 with > 20–30%, 36 with > 30–40%, 18 with > 40–50%, and 21 with > 50%. Of the sites including development features, 22 sites contained permanent polygonal features, 36 with permanent linear, 44 with well sites, 108 with wide linear features, and 165 with seismic. Approximately a third of the sites were within each distance range of 0–0.5 km, 0.5–2 km, and > 2 km from permanent features (103, 88, and 93 sites respectively). To account for potential confounding effects between the sampling schedule (time-of-day) and the spatial location of development features within the study area, the start points for daily survey routes were randomly selected across different distances from permanent features and otherwise randomly distributed in space to account for any unknown confounding effects of sampling date and location. Point-count surveys were conducted once at each sampling site, prioritizing the importance of a larger sample size (Ralph et al. 1993) to examine a greater range of development-feature combinations.
Skilled observers visited a total of 284 sites during the breeding season between 4 June and 1 July 2014. At each location, observers recorded songbirds seen and heard within a 100-m, limited-radius sample area and over a 10-minute sampling duration. Surveys occurred during good weather (no rain, light wind) and during peak hours of avian activity between approximately 04:00 and 9:45. We focused our analysis on 157 of these sites, which (1) contained a minimum of 20% lowland habitat within 100 m of the survey site; and (2) were within the extent of available high-resolution habitat data (Fig. 2). The reduced sample set contained a similar average and range for amount of development for each feature type within 100 m of sample stations relative to the full dataset; however, we had a slightly higher mean total development feature amount (25.5 ± 17% relative to 22.0 ± 19%) because of a reduction in the number of sites farthest from permanent roads. See Appendix 2 for additional detail on development-feature distribution within the sample dataset.
Habitat variables for this study were derived from three sources: (1) an avian habitat class layer derived from human-classified aerial imagery (Mahon et al. 2016); (2) 2009 Light Detection and Ranging (LiDAR) data (bare ground and full-feature light detection and ranging data); and (3) 2013 Pleiades 50-cm resolution multispectral satellite imagery. Broad habitat types were identified from the avian habitat class layer which distinguishes vegetation into 12 types based on stand-level vegetation associations, such as leading species composition, and includes up to six different structural stage classes for each habitat type. We derived finer-resolution vegetation-structure characteristics, including vegetation density and height, using LiDAR data.
Vegetation productivity has been correlated with avian abundance and richness for a number of different forest-dominated systems (Hurlbert and Haskell 2003, Evans et al. 2006, St-Louis et al. 2014). Here we identified site productivity as the average of the normalized difference vegetation index (NDVI) of lowland vegetation within 100 m of the survey site. NDVI indices, which are calculated using a ratio of near infrared (NIR) and visible (VIS) wavelength spectral reflectance from satellite imagery, NDVI = (NIR-VIS) / (NIR+VIS), represent areas of high chlorophyll concentrations and are thought to correspond to differences in vegetation productivity or so-called greenness (Pettorelli et al. 2011). To aid interpretability, we scaled our vegetation productivity or greeness index by 1000 NDVI, to represent relative change for values greater than 1, instead of increments of 0.001.
Defining development features
A combination of automated and manual digitization techniques were used to delineate development features from both planning schematics and Pleaides 50-cm resolution satellite imagery (September 2013). As new vegetation clearing occurs during the winter months, it is possible that some new features were present on the ground during the time of the point-count surveys. However, development data from May 2015 (Pleaides 50-cm resolution satellite imagery) showed that only two sites had ≥ 1% change in development feature area within 100 m of the point count station (3% and 14% respectively) and neither of these sites were outliers in fitted models. Development features were grouped into five distinct categories: (1) permanent polygonal features, such as gravel pits, developed well sites, and other industrial features; (2) permanent linear features such as gravel roads (25–70 m wide); (3) temporary well sites such as undeveloped exploratory well sites (approximately 50 × 80 m polygons, hereafter well sites); (4) temporary wide linear features such as traditional linear cut-line features, pipelines, and winter roads (5–20 m wide); and (5) temporary seismic features, limited to modern 2–3 m wide 3D linear seismic features (hereafter seismic; Fig. 1). To identify whether different types of development features have different effects on Dark-eyed Junco and Palm Warbler occurrence, we examined the percent area of different combinations of development-feature types (Table 1).
Model selection and analysis
As evidence suggests avian species respond to different factors at different scales (Leonard et al. 2008) and that multi-scale models improve the predictive capacity of habitat selection models (Smith et al. 2011), we considered habitat and development variables at two spatial scales using circular buffers around the point-count station (Fig. 3). We selected a 500-m radius neighborhood scale to (1) be large enough to contain home ranges of multiple individuals, thus encompassing influential biological processes such as dispersal constraints and conspecific attraction (Desrochers et al. 2010); (2) fall within a range considered influential in the literature for songbirds (Desrochers et al. 2010); and (3) capture a representative gradient of composition across habitats and target development-feature types. We selected a 100-m local scale to (1) match the zone of detection from the limited-radius sampling method, thus capturing local-scale differences in habitat around each station; and (2) represent local-scale processes such as nest-site availability, microclimate, or food limitation (Desrochers et al. 2010, Farwell et al. 2016). For each habitat or disturbance variable, scale was evaluated first, and the most predictive scale selected for inclusion in multi-scale models (Appendix 1). We considered the most predictive scale the one with the lowest Akaike’s Information Criterion (AIC) score when comparing single-variable models with different scales.
Once an appropriate scale was established for each variable, a two-step approach was used to compare model hypotheses. First, we developed a habitat only (null) model for each species (Appendix 1). The habitat model was developed using a limited number of a priori predictor variables representing three habitat characteristics: habitat amount, vegetation structure, and vegetation productivity. To identify the best supported variable for habitat amount, we examined combinations of lowland categories with different vegetation composition and structural stage attributes, characteristics important to forest-associated boreal birds (Machtans and Latour 2003, Schieck and Song 2006, Mahon et al. 2016). Categories included (1) lowland (all stages of bog and fen combined); (2) shrubby bog (≤ 6 m tall vegetation); (3) shrubby fen; (4) treed bog (≥ 6 m tall vegetation); (5) treed fen; (6) shrubby lowland (bog and fen combined); and (7) treed lowland (bog and fen combined). After identifying the best predictor for habitat amount, we then determined if local vegetation structure and vegetation productivity improved the predictive capacity of models. Variables considered for vegetation structure included density and variability of lowland vegetation structure. Vegetation density was represented by percent area of preferred shrubby (1 to < 4 m tall) or regenerating (4 to < 10 m tall) vegetation within lowland habitat. Variability in vegetation structure was represented by the standard deviation of vegetation heights within lowland vegetation. To avoid overfitting, only the single best predictor for each variable category (habitat amount, vegetation structure, vegetation productivity) was included in the final habitat model for each species, and vegetation structure or productivity variables were only included if they improved the model above only habitat amount.
Next, different combinations of development features were added to the habitat (null) models to compare relative influence of different features. Because the majority of survey stations had ≤ 1 detection for each species (133/157 stations for juncos and 139/157 for Palm Warblers), we examined the relative influence of predictor variables on whether or not a species was detected at each site. Logistic regression models were used to model the probability of occurrence (1 = detected, 0 = not detected) for each species as a function of percent area of each feature type, scaled to show the change in probability of occurrence per 10% change in the explanatory feature. We assessed all models for variable collinearity using variance inflation factors (all VIF < 3; Zuur et al. 2010). Development feature models were ranked by their ability to predict songbird occurrence relative to the null habitat quality model for each species using Akaike Information Criterion corrected for small sample sizes (AICc; Burnham and Anderson 2002). Models with the lowest AICc and greatest Akaike weights (wi) were considered the most parsimonious and selected for each species. Coefficients were standardized to a variance of 1 and mean of 0 to compare relative influence for variables with different measures. Strong positive or negative coefficients suggest correlations between development features and occurrence. Predictive accuracy of models was evaluated using the receiver operating characteristic area under the curve (ROC AUC). Analyses were completed in R (R Core Team 2020) and were facilitated using the packages lme4 (Bates et al. 2018), MuMIn (Bartoń 2019), and pROC (Robin et al. 2011).
Occurrence and development
A total of 107 Dark-eyed Juncos and 96 Palm Warblers (singing males only) were detected within a 100-m sampling radius of point-count centers for the 157 point-count stations analyzed. Dark-eyed Juncos occurred at 78 of these stations and Palm Warblers at 73. Percent lowland within 100 m of stations ranged from 20.7–100% (mean 68.1 ± 19.1) at the 100-m scale and 19.7–96.9% at the 500 m scale (mean 63.8 ± 16.7; Table 1). Similarly, sampling stations contained comparable proportions of each different development feature type at both sample scales (Table 1). For all feature types combined, there was an average of 25.8% (± 17.3, 0–72.7%) and 24.9% (± 14.2, 2.1–74.8%) area of development within 100 m and 500 m respectively. A total of 28 point-count stations contained less than 5% development within 100 m, whereas only 14 stations contained less than 5% development within 500 m.
Habitat null model
Inclusion of vegetation structure and productivity variables substantially improved model fit for Palm Warblers, but not for Dark-eyed Juncos. For Juncos, the top model included percent area of treed bog within 500 m of point-count station, but the model was not improved by inclusion of additional local-scale variables of vegetation structure or productivity (Table 2; wi: 0.04). Junco occurrence increased 1.4 times for every 10% increase in treed bog habitat (Fig. 4, Table 3). For Palm Warblers, the top habitat model included percent area of shrubby lowland habitat (bog or fen) within 500 m of the point-count; however, there was similar model fit at both the 500 m and 100 m scales (< Δ2 AIC; see Table A1.3 in Appendix 1). The habitat model was improved by the inclusion of local scale variables of both lowland structure and vegetation productivity (Table 4; wi: < 0.01). Palm Warbler occurrence was 1.2 times more likely for every 10% increase in the amount of shrubby lowland habitat, while being 0.37 times as likely for every 1 m increase in the standard deviation of tree height and 1.27 times more likely for every increase in average lowland greenness by 1 (0.001 NDVI; Fig. 4, Table 5). In our sampled locations, the scaled relative greenness of lowland vegetation ranged from 56.3 to 70.3.
Feature-specific response to permanent development
The most supported model for both Dark-eyed Junco and Palm Warbler occurrence included only permanent features. The top Dark-eyed Junco model (Table 2; wi: 0.64) had a ROC AUC of 0.62, indicating weak predictive ability. The low sensitivity (0.47) relative to specificity (0.72) of the model suggests that Dark-eyed Juncos occur in more locations than the model predicted. For every 10% increase in the percent area of polygonal permanent features the odds of Dark-eyed Junco occurrence decreased by 0.6 times, whereas the odds of occurrence increased 1.7 times for every 10% increase in permanent linear features (Table 3, Fig. 5). Percent area of linear permanent features within 100 m had the greatest relative effect on occurrence based on standardized beta coefficients, followed by the percent area of permanent polygonal features within 500 m, and then percent area of treed bog within the 500-m neighborhood scale (Table 3). Only two other models showed more support than the habitat only (null) model, and both these models contained permanent features (linear only wi: 0.14 and polygonal only wi: 0.08; Table 2). All models containing only temporary development features ranked well below the null habitat-only model based on AICc (Table 2; wi: 0.03).
For Palm Warblers, the top model (Table 4, wi: 0.63) was supported by a ROC AUC of 0.78, with similar specificity (0.71) and sensitivity (0.68), indicating good model accuracy. The most to least influential variables based on standardized beta coefficients were percent area of polygonal permanent features at a local scale, local tree height variability, local NDVI, neighborhood percent area shrubby lowland, and neighborhood percent area linear permanent features (Table 5). Odds of occurrence decreased 0.4 times for every 10% increase in permanent polygonal features within a 100-m local scale and increased 3.9 times for every 10% increase in area of permanent linear features within a 500-m neighborhood (Figure 5, Table 5). All models including permanent development features ranked better than the habitat quality null model (Table 4).
In situ oil-sands development features are expected to become increasingly prevalent in northern Alberta boreal forests (Schneider and Dyer 2006). Understanding species’ sensitivity to different types of development features is important for predicting and managing responses to expanding anthropogenic footprints. Our results indicate that both Dark-eyed Juncos and Palm Warblers are influenced by the presence of permanent development features, with both species showing a negative response to polygonal permanent features and a positive response to permanent linear features. In contrast, despite differences in specificity of habitat requirements, both species are commonly present within areas perforated by smaller temporary development features, supporting the hypothesis that temporary features are lower-impact than more permanent features.
Responses to permanent features
Permanent features showed consistent effects on the occurrence of Dark-eyed Juncos and Palm Warblers, despite differences between the species in their broad-scale habitat preference. Inclusion of polygonal and linear permanent features improved model fit for both species, supporting the hypothesis that permanent features are not equivalent to natural habitats and have a greater influence on habitat suitability than temporary development features. Polygonal permanent features had a strong negative influence on the probability of occurrence. Polygonal permanent features here include developed SAGD well sites, gravel pits, and industrial facilities. Associated human activity, industrial noise, movement-impeding infrastructure, and high contrast to the surrounding habitat are some potential reasons why these permanent polygonal features provide less suitable habitat.
Contrary to our predictions, both species showed a positive response to permanent linear features and the relative strengths of responses were similar to those of larger polygonal features. In this study area, all permanent linear features were unpaved gravel roads used to access industrial site locations. In forested ecosystems, responses to unpaved roads on abundance or density of passerines is quite variable (Ortega and Capen 2002). Road networks affect bird populations through different direct and indirect mechanisms. Direct negative impacts may include habitat loss, mortality from vehicle collisions, or poisoning from roadside pollution, whereas indirect effects may include influences of artificial light or noise on breeding success, physical barriers to movement, or edge effects (Kociolek et al. 2011). Positive influences on passerines may include provision of novel early successional habitat in the verge and potentially in the forest edge, resulting in increased landscape heterogeneity (Helldin and Seiler 2003) and altered distribution or availability of food resources (Morelli et al. 2014).
Ortega and Capen (2002) reported a positive response to roads for Dark-eyed Juncos, observing increased abundance of juncos within 150 m of unpaved roads relative to forest interiors. They attributed this response to low herbaceous and woody vegetation along roadsides because of observations of juncos foraging within low vegetation and directly on unpaved road surfaces. In our study, juncos were also observed foraging both on gravel substrates and along roadside edges (Carpenter 2020), so gravel roads may influence availability or diversity of insect-prey species for ground foragers. Although, as foliage gleaners, Palm Warblers do not commonly feed within herbaceous vegetation or along the ground, they might benefit from changes in insect populations along the interface of gravel roads and surrounding peatland vegetation. Altered sunlight or moisture along roadside edges can increase the abundance or diversity of insect-prey species within the surrounding vegetation, leading to shifts in the insect community that may impact predator-prey dynamics for avian species (Muñoz et al. 2015, Riva et al. 2018b). Conversely, aerial-insect abundance may decline with increasing road traffic and the abundance of some species can be lower closer to roadside edges (Muñoz et al. 2015, Martin et al. 2018), so positive responses could be limited to low-traffic gravel roads similar to those in this study.
Roads in peatlands can have variable effects on surrounding vegetation depending on the type of peatland and road orientation. When roads intersect peatlands at an orientation perpendicular to the direction of the underlying hydrological flow they may act as water barriers (Willier 2017). The upstream side of the road can become saturated, leading to widescale vegetation mortality, whereas drier conditions on the downstream side of the road may enhance woody vegetation growth, resulting in forest-like conditions. The road networks in this study area are relatively recent and intersect the peatland habitat at various orientations. Vegetation change has been observed in some locations adjacent to the road network that are suggestive of the potential for edge effects on vegetation from road-based constraints to hydrologic flow in peatland ecosystems. Post hoc analyses suggested model fit for Palm Warblers was somewhat improved by a non-significant positive-interaction effect between peatland productivity (NDVI) and permanent linear features. Because vegetation productivity in peatlands will differ between bogs and fens, which have different hydrological characteristics and thus will show different responses to road disturbances (Willier 2017), this may be worth further exploring. Additional investigation into the mechanism behind these responses, the consistency of this response over time, and the consistency of positive responses for other passerine species, or in other peatland areas, is needed.
In general, this work supports other findings that impacts from roads differ from cutlines (wide linear or 3D seismic features) and cannot be directly extrapolated (Linke et al. 2008). Whereas cutlines create greater edge per unit area than other features (Linke et al. 2008, Bayne et al. 2016, Riva and Nielsen 2020), they may also create softer edges that are not perceived by these two species as structurally distinct to natural openings (Machtans 2006). Additionally, factors including resilience and threshold effects may contribute to the relatively weak influence of temporary development features on Dark-eyed Junco and Palm Warbler occurrence.
Resilience or threshold effects
The boreal forest is a naturally heterogeneous ecosystem which may help explain why small temporary development features did not affect the occurrence of either study species. The high frequency of natural disturbance events, including fire and insect outbreaks, creates a patchy mosaic of upland and lowland forest types in various stages of succession (Bergeron et al. 2014). As a result, species that evolved in dynamic boreal forest may be well-adapted to heterogeneous vegetation structure, and thus fairly resilient to small changes in vegetation structure from disturbances (Schmiegelow et al. 1997, Drapeau et al. 2016, Mahon et al. 2016). Vegetation structure in peatland habitats is closely tied to underlying patterns of nutrient and moisture regimes and naturally includes small vegetation gaps and variation in tree height, although perhaps at a finer scale than in neighboring upland habitats. The low nutrient availability and high moisture regime may limit overall heterogeneity in species composition, and corresponding vegetation structure creating more naturally variegated habitats (Harper et al. 2015). With a lower tree density and wider spacing, peatland habitats may be affected less by small structural changes from temporary disturbance features.
Moreover, fragmentation effects, such as patch isolation, may be stronger in highly disturbed landscapes, such as urban or agricultural areas, where habitat availability is low and the matrix imposes greater survival risk during movement (Betts et al. 2010, Villard and Metzger 2014). In the boreal forest region, energy-sector developments may act more like within-patch perforation of larger contiguous forest than factors inducing patch-isolation effects. In hardwood and oak ecosystems, avian communities at sites with low well-site densities comparable to this study area (4-20 well sites per km²) were similar to reference sites, but these communities diverged at higher development densities (> 20 well sites per km²), suggesting a threshold effect. Ovenbird responses to conventional seismic lines were also similar, with no change detected between 0-8.6 km of conventional seismic per km² but 19% declines for each km per km² above that threshold (Bayne et al. 2005b). Development feature densities for temporary wide linear features in this study area were 0–4.6 km per km² and may be below thresholds for strong Dark-eyed Junco and Palm Warbler responses.
Generalist and specialist responses
Despite similar directions of response to permanent features by these two species, these findings add to existing evidence that specialized species with a narrower ecological niche breadth may be more susceptible to habitat change than species that use a broader range of habitat characteristics. For Palm Warblers, their occurrence in lowland habitats increased with shorter vegetation heights, lower variability in vegetation height, and greater vegetation productivity, suggesting that they select areas with a greater aggregation of shrubby vegetation strata. This is consistent with observations of Palm Warblers using short black spruce trees (< 4 m) for both singing and foraging behaviors in this study area (Carpenter 2020). Additionally, standardized coefficients suggest that vegetation productivity has the greatest positive effect on Palm Warbler occurrence. Conservation and management of this species therefore requires careful consideration of the structural integrity and productivity of these habitats. If this habitat specificity holds true across their range, it may explain why this species is considered area-sensitive in eastern boreal forests (Poulin et al. 2006). In this study area, lowland habitats were widespread and fairly contiguous, with large mean patch sizes relative to other habitat types, so area sensitivity of Palm Warblers may not be apparent under these conditions.
The relatively poor model accuracy for Dark-eyed Juncos indicates that juncos are commonly present in areas where models do not predict they would be, suggesting generalist habitat use. Whereas our models could be missing covariates that better explain Dark-eyed Junco habitat use, this result is not unexpected. Most species in the boreal forest exhibit generalist rather than specialist tendencies (Schieck and Song 2006, Mahon et al. 2016), and this flexibility in habitat use makes it difficult to identify the key factors influencing habitat suitability. This work demonstrates that predicting responses to energy development for generalist species is likely to be complex and challenging.
A consideration for interpreting the response to development features in our study is that we used single-visit occurrence data that was not corrected for detectability. Avian detectability is known to be influenced by factors including ambient noise (e.g., wind or anthropogenic), vegetation structure (e.g., forested or open habitat types), bird behavior (e.g., song frequency, vocalization spectrum), distance from the observer, and study or analytical design (e.g., point-count duration, detection radius; Alldredge et al. 2007, Sólymos et al. 2013, 2018, Hutto 2016, Yip et al. 2017). Whereas we aimed to minimize detectability bias in our choice of point count method, e.g., 10-minute, limited radius, in good survey conditions, with experienced observers (Matsuoka et al. 2014) and by using a stratified approach to sampling design (e.g., sampling across a range of distances to roadsides and a gradient of development feature amounts), we acknowledge that imperfect detection is an inherent aspect of our data. One potential implication here is that larger features may be associated with more open structure, and therefore increased detectability, if the observer is within or adjacent to the opening (Yip et al. 2017). In contrast, any anthropogenic noise associated with permanent features could decrease detectability. Whereas road traffic within our study area was low during the survey period, human activity within polygonal features containing associated infrastructure was greater than on access roads, suggesting a bias toward detection along roadsides relative to active worksites; however, not all polygonal features contained infrastructure at the time of the surveys. Therefore, whereas this study supports inference toward the relative effect of these features, additional work is needed to quantify the magnitude of the effects of permanent features on species’ occupancy and density.
These findings emphasize the importance of minimizing impacts from widespread permanent features associated with in situ energy developments. As in situ oil-sands development expands in northern Alberta, so will densities of associated permanent infrastructure, including gravel or paved roads, SAGD well sites, and facilities. Moreover, these impacts will not be occurring in isolation, but in conjunction with other industries such as transportation and mining (Ficken et al. 2019). Impacts from different development features (roads, narrow and wide linear features, well sites, harvest units) in the boreal forest region can show additive or interactive effects (Mahon et al. 2019). Because even temporary features will persist, un-regenerated, for extended periods in peatland habitats relative to uplands (Lee and Boutin 2006, Van Rensen et al. 2015), there is greater uncertainty in the long-term responses for species associated with these habitats. It will become increasingly important to consider forward planning to minimize forest alteration and integrate land use across industries and lease holders. Additional work at more lowland sites impacted by energy development would help assess the generality of these patterns.
These findings also support the suggestion that habitat amount has a stronger effect than fine-scale fragmentation (Fahrig 2003), while still showing important effects of different permanent development features on surrounding habitat. This study, however, does not address the effects of small-scale permanent and temporary development features on habitat quality, or evaluate potential behavioral implications for species that occur in these developed habitats. Whereas evaluating species occurrence can help identify the relative importance of different development features, it does not capture changes in species density or reproductive success. Exploring these other important measures may reveal additional or more subtle impacts for some development feature types. More work on the effects of processing facilities, compressor stations, and other more permanent development features is also needed, especially in peatlands. We recommend additional, more detailed studies to address these outstanding questions in this poorly studied vegetation type.
RESPONSES TO THIS ARTICLEResponses to this article are invited. If accepted for publication, your response will be hyperlinked to the article. To submit a response, follow this link. To read responses already accepted, follow this link.
C. L. M. and J. L. K. conceptualized the project, including designing field data collection, and provided supervision. T. M. C. conducted the analyses and wrote the manuscript. All other authors provided analytical guidance and contributed review and editing feedback to development of the manuscript.
We thank J. Bahr, J. L. Keim, and T. Mahon for collecting the data used in this project. C. L. Mahon, J. L. Keim, and C. Robinson organized the collaboration that made this project possible. We are grateful to support from C. Robinson as well as the Environmental Team at Brion Energy who facilitated on-site activities. We received fantastic Geological Information System expertise from F. Wyatt and G. Holloway of Fiera Biological Consulting and ongoing project support from S.J. Song at Environment and Climate Change Canada (ECCC). This manuscript benefitted from valuable feedback by J. Toms and J. Ball. Funding for this project was provided by the Oil Sands Monitoring (OSM) Program, ECCC, Matrix Environmental Solutions, and Brion Energy. This work is independent of any position of the OSM Program.
Alberta Biodiversity Monitoring Institute (ABMI). 2017. The status of human footprint in Alberta. ABMI, University of Alberta, Edmonton, Alberta, Canada.
Alberta Government. 2017. Oil sands facts and stats. Alberta Government, Edmonton, Alberta, Canada.
Alberta Government. 2020. Annual report energy 2019–2020. Alberta Government, Energy Communications, Edmonton, Alberta, Canada.
Alldredge, M. W., T. R. Simons, and K. H. Pollock. 2007. Factors affecting aural detections of songbirds. Ecological Applications 17:948-955. https://doi.org/10.1890/06-0685
Bartoń, K. 2019. MuMIn: Multi-Model Inference, R Package Version 1.43.15. https://CRAN.R-project.org/package=MuMIn
Bates, D., M. Maechler, B. Bolker, and S. Walker. 2018. lme4: Linear Mixed-Effects Models Using “Eigen” and S4, R Package Version 1.1-30. https://cran.r-project.org/web/packages/lme4/index.html
Bayne, E. M., S. Boutin, B. Tracz, and K. Charest. 2005a. Functional and numerical responses of Ovenbirds (Seiurus aurocapilla) to changing seismic exploration practices in Alberta’s boreal forest. Écoscience 12:216-222. https://doi.org/10.2980/i1195-6860-12-2-216.1
Bayne, E. M., L. Leston, C. L. Mahon, P. Sólymos, C. Machtans, H. Lankau, J. R. Ball, S. L. Van Wilgenburg, S. G. Cumming, T. Fontaine, et al. 2016. Boreal bird abundance estimates within different energy sector disturbances vary with point count radius. Condor 118:376-390. https://doi.org/10.1650/CONDOR-15-126.1
Bayne, E. M., S. L. Van Wilgenburg, S. Boutin, and K. A. Hobson. 2005b. Modeling and field-testing of Ovenbird (Seiurus aurocapillus) responses to boreal forest dissection by energy sector development at multiple spatial scales. Landscape Ecology 20:203-216. https://doi.org/10.1007/s10980-004-2265-9
Beckingham, J. D., and J. H. Archibald. 1996. Field guide to ecosites of Northern Alberta. Natural Resources Canada, Canadian Forest Service, Northern Forestry Centre, Edmonton, Alberta, Canada.
Bergeron, Y., H. Y. H. Chen, N. C. Kenkel, A. L. Leduc, and S. E. Macdonald. 2014. Boreal mixedwood stand dynamics: ecological processes underlying multiple pathways. Forestry Chronicle 90:202-213. https://doi.org/10.5558/tfc2014-039
Bernath-Plaisted, J., and N. Koper. 2016. Physical footprint of oil and gas infrastructure, not anthropogenic noise, reduces nesting success of some grassland songbirds. Biological Conservation 204:434-441. https://doi.org/10.1016/j.biocon.2016.11.002
Betts, M. G., L. Fahrig, A. S. Hadley, K. E. Halstead, J. Bowman, W. D. Robinson, J. A. Wiens, and D. B. Lindenmayer. 2014. A species-centered approach for uncovering generalities in organism responses to habitat loss and fragmentation. Ecography 37:517-527. https://doi.org/10.1111/ecog.00740
Betts, M. G., J. C. Hagar, J. W. Rivers, J. D. Alexander, K. McGarigal, and B. C. McComb. 2010. Thresholds in forest bird occurrence as a function of the amount of early-seral broadleaf forest at landscape scales. Ecological Applications 20:2116-2130. https://doi.org/10.1890/09-1305.1
Brant, J. P. 2009. The extent of the North American boreal zone. Environmental Reviews 17:101-161. https://doi.org/10.1139/A09-004
Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second edition. Springer New York, New York, USA.
Calmé, S., and A. Desrochers. 2000. Biogeographic aspects of the distribution of bird species breeding in Québec’s peatlands. Journal of Biogeography 27:725-732. https://doi.org/10.1046/j.1365-2699.2000.00434.x
Calmé, S., A. Desrochers, and J.-P. L. Savard. 2002. Regional significance of peatlands for avifaunal diversity in southern Québec. Biological Conservation 107:273-281. https://doi.org/10.1016/S0006-3207(02)00063-0
Carpenter, T. M. 2020. Effects of energy development on habitat use of an avian peatland specialist and generalist at multiple spatial scales. Thesis. University of Alberta, Edmonton, Alberta, Canada.
Carrara, E., V. Arroyo-Rodríguez, J. H. Vega-Rivera, J. E. Schondube, S. M. de Freitas, and L. Fahrig. 2015. Impact of landscape composition and configuration on forest specialist and generalist bird species in the fragmented Lacandona rainforest, Mexico. Biological Conservation 184:117-126. https://doi.org/10.1016/j.biocon.2015.01.014
Clavel, J., R. Julliard, and V. Devictor. 2011. Worldwide decline of specialist species: toward a global functional homogenization? Frontiers in Ecology and the Environment 9:222-228. https://doi.org/10.1890/080216
Delgado, J. D., N. L. Arroyo, J. R. Arévalo, and J. M. Fernández-Palacios. 2007. Edge effects of roads on temperature, light, canopy cover, and canopy height in laurel and pine forests (Tenerife, Canary Islands). Landscape and Urban Planning 81:328-340. https://doi.org/10.1016/j.landurbplan.2007.01.005
Desrochers, A., C. Renaud, W. M. Hochachka, and M. Cadman. 2010. Area-sensitivity by forest songbirds: theoretical and practical implications of scale-dependency. Ecography 33:921-931. https://doi.org/10.1111/j.1600-0587.2009.06061.x
Dickie, M., R. Serrouya, R. S. McNay, and S. Boutin. 2017. Faster and farther: wolf movement on linear features and implications for hunting behaviour. Journal of Applied Ecology 54:253-263. https://doi.org/10.1111/1365-2664.12732
Drapeau, P., M. A. Villard, A. Leduc, and S. J. Hannon. 2016. Natural disturbance regimes as templates for the response of bird species assemblages to contemporary forest management. Diversity and Distributions 22:385-399. https://doi.org/10.1111/ddi.12407
Evans, K. L., N. A. James, and K. J. Gaston. 2006. Abundance, species richness, and energy availability in the North American avifauna. Global Ecology and Biogeography 15:372-385. https://doi.org/10.1111/j.1466-822X.2006.00228.x
Ewers, R. M., and R. K. Didham. 2006. Continuous response functions for quantifying the strength of edge effects. Journal of Applied Ecology 43:527-536. https://doi.org/10.1111/j.1365-2664.2006.01151.x
Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology, Evolution, and Systematics 34:487-515. https://doi.org/10.1146/annurev.ecolsys.34.011802.132419
Farwell, L. S., P. B. Wood, J. Sheehan, and G. A. George. 2016. Shale gas development effects on the songbird community in a central Appalachian forest. Biological Conservation 201:78-91. https://doi.org/10.1016/j.biocon.2016.06.019
Ficken, C. D., D. Cobbaert, R. C. Rooney. 2019. Low extent but high impact of human land use on wetland flora across the boreal oil sands region. Science of the Total Environment 693:133647. https://doi.org/10.1016/j.scitotenv.2019.133647
Fischer, J., and D. B. Lindenmayer. 2007. Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography 16:265-280. https://doi.org/10.1111/j.1466-8238.2007.00287.x
Gillies, C. S., and C. C. St. Clair. 2010. Functional responses in habitat selection by tropical birds moving through fragmented forest. Journal of Applied Ecology 47:182-190. https://doi.org/10.1111/j.1365-2664.2009.01756.x
Handel, C. M., S. A. Swanson, D. A. Nigro, and S. M. Matsuoka. 2009. Estimation of avian population sizes and species richness across a boreal landscape in Alaska. Wilson Journal of Ornithology 121:528-547. https://doi.org/10.1676/08-067.1
Harper, K. A., S. E. Macdonald, M. S. Mayerhofer, S. R. Biswas, P. A. Esseen, K. Hylander, K. J. Stewart, A. U. Mallik, P. Drapeau, B. G. Jonsson, et al. 2015. Edge influence on vegetation at natural and anthropogenic edges of boreal forests in Canada and Fennoscandia. Journal of Ecology 103:550-562. https://doi.org/10.1111/1365-2745.12398
Helldin, J.-O., and A. Seiler. 2003. Effects of roads on the abundance of birds in Swedish forest and farmland. Pages 1-9 in Proceedings of the IENE Conference on Habitat Fragmentation Due to Transportation Infrastructure. Belgium Institute for Nature Conservation, Brussels, Belgium.
Hurlbert, A. H., and J. P. Haskell. 2003. The effect of energy and seasonality on avian species richness and community composition. American Naturalist 161:83-97. https://doi.org/10.1086/345459
Hutto, R. L. 2016. Should scientists be required to use a model-based solution to adjust for possible distance-based detectability bias? Ecological Applications 26:1287-1294. https://doi.org/10.1002/eap.1385
Jordaan, S. M., D. W. Keith, and B. Stelfox. 2009. Quantifying land use of oil sands production: a life cycle perspective. Environmental Research Letters 4:024004. https://doi.org/10.1088/1748-9326/4/2/024004
Kennedy, C. M., P. P. Marra, W. F. Fagan, and M. C. Neel. 2010. Landscape matrix and species traits mediate responses of Neotropical resident birds to forest fragmentation in Jamaica. Ecological Monographs 80:651-669. https://doi.org/10.1890/09-0904.1
Kociolek, A. V., A. P. Clevenger, C. C. St. Clair, and D. S. Proppe. 2011. Effects of road networks on bird populations. Conservation Biology 25:241-249. https://doi.org/10.1111/j.1523-1739.2010.01635.x
Lankau, H. E., E. M. Bayne, and C. S. Machtans. 2013. Ovenbird (Seiurus aurocapilla) territory placement near seismic lines is influenced by forest regeneration and conspecific density. Avian Conservation and Ecology 8:5. https://doi.org/10.5751/ACE-00596-080105
Lee, P., and S. Boutin. 2006. Persistence and developmental transition of wide seismic lines in the western Boreal Plains of Canada. Journal of Environmental Management 78:240-250. https://doi.org/10.1016/j.jenvman.2005.03.016
Leonard, T. D., P. D. Taylor, and I. G. Warkentin. 2008. Landscape structure and spatial scale affect space use by songbirds in naturally patchy and harvested boreal forests. Condor 110:467-481. https://doi.org/10.1525/cond.2008.8512
Linke, J., S. E. Franklin, M. Hall-Beyer, and G. B. Stenhouse. 2008. Effects of cutline density and land-cover heterogeneity on landscape metrics in western Alberta. Canadian Journal of Remote Sensing 34:390-404. https://doi.org/10.5589/m08-034
Ludlow, S. M., R. M. Brigham, and S. K. Davis. 2015. Oil and natural gas development has mixed effects on the density and reproductive success of grassland songbirds. Condor 117:64-75. https://doi.org/10.1650/CONDOR-14-79.1
MacFarlane, A. K. 2003. Vegetation responses to seismic lines: edge effects and on-line succession. Thesis. University of Alberta, Edmonton, Alberta, Canada.
Machtans, C. S. 2006. Songbird response to seismic lines in the western boreal forest: a manipulative experiment. Canadian Journal of Zoology 84:1421-1430. https://doi.org/10.1139/z06-134
Machtans, C. S., and P. B. Latour. 2003. Boreal forest songbird communities in the Liard Valley, Northwest Territories, Canada. Condor 105:27-44. https://doi.org/10.1093/condor/105.1.27
Mahon, C. L., G. L. Holloway, E. M. Bayne, and J. D. Toms. 2019. Additive and interactive cumulative effects on boreal landbirds: winners and losers in a multi-stressor landscape. Ecological Applications 29:1-18. https://doi.org/10.1002/eap.1895
Mahon, C. L., G. Holloway, P. Sólymos, S. G. Cumming, E. M. Bayne, F. K. A. Schmiegelow, and S. J. Song. 2016. Community structure and niche characteristics of upland and lowland western boreal birds at multiple spatial scales. Forest Ecology and Management 361:99-116. https://doi.org/10.1016/j.foreco.2015.11.007
Martin, A. E., S. L. Graham, M. Henry, E. Pervin, and L. Fahrig. 2018. Flying insect abundance declines with increasing road traffic. Insect Conservation and Diversity 11:608-613. https://doi.org/10.1111/icad.12300
Matsuoka, S. M., C. L. Mahon, C. M. Handel, P. Sólymos, E. M. Bayne, P. C. Fontaine, and C. J. Ralph. 2014. Reviving common standards in point-count surveys for broad inference across studies. Condor 116:599-608. https://doi.org/10.1650/CONDOR-14-108.1
Morelli, F., M. Beim, L. Jerzak, D. Jones, and P. Tryjanowski. 2014. Can roads, railways, and related structures have positive effects on birds? - a review. Transportation Research Part D: Transport and Environment 30:21-31. https://doi.org/10.1016/j.trd.2014.05.006
Morissette, J. L., K. J. Kardynal, E. M. Bayne, and K. A. Hobson. 2013. Comparing bird community composition among boreal wetlands: is wetland classification a missing piece of the habitat puzzle? Wetlands 33:653-665. https://doi.org/10.1007/s13157-013-0421-1
Mumma, M. A., M. P. Gillingham, C. J. Johnson, and K. L. Parker. 2017. Understanding predation risk and individual variation in risk avoidance for threatened boreal caribou. Ecology and Evolution 7:10266-10277. https://doi.org/10.1002/ece3.3563
Muñoz, P. T., F. P. Torres, and A. G. Megías. 2015. Effects of roads on insects: a review. Biodiversity and Conservation 24:659-682. https://doi.org/10.1007/s10531-014-0831-2
Northrup, J. M., and G. Wittemyer. 2013. Characterising the impacts of emerging energy development on wildlife, with an eye towards mitigation. Ecology Letters 16:112-125. https://doi.org/10.1111/ele.12009
Ortega, Y. K., and D. E. Capen. 2002. Roads as edges: effects on birds in forested landscapes. Forest Science 48:381-390.
Pattison, C. A., M. S. Quinn, P. Dale, and C. P. Catterall. 2016. The landscape impact of linear seismic clearings for oil and gas development in boreal forest. Northwest Science 90:340. https://doi.org/10.3955/046.090.0312
Pettorelli, N., S. Ryan, T. Mueller, N. Bunnefeld, B. Jedrzejewska, M. Lima, and K. Kausrud. 2011. The Normalized Difference Vegetation Index (NDVI): unforeseen successes in animal ecology. Climate Research 46:15-27. https://doi.org/10.3354/cr00936
Pickell, P., S. Gergel, N. Coops, and D. Andison. 2014. Monitoring forest change in landscapes under-going rapid energy development: challenges and new perspectives. Land 3:617-638. https://doi.org/10.3390/land3030617
Poulin, M., M. Bélisle, and M. Cabeza. 2006. Within-site habitat configuration in reserve design: a case study with a peatland bird. Biological Conservation 128:55-66. https://doi.org/10.1016/j.biocon.2005.09.016
Prevedello, J. A., and M. V. Vieira. 2010. Does the type of matrix matter? A quantitative review of the evidence. Biodiversity and Conservation 19:1205-1223. https://doi.org/10.1007/s10531-009-9750-z
Ralph, C. J., S. Droege, and J. R. Sauer. 1995. Managing and monitoring birds using point counts: standards and applications. Pages 161-175 in C. J. Ralph, S. Droege, and J. R. Sauer, editors. Monitoring bird populations by point counts. U.S. Department of Agriculture, Forest Service, Pacific Southwest Research Station, Albany, California, USA. https://doi.org/10.2737/PSW-GTR-149
Ralph, C. J., G. R. Geupel, P. Pyle, T. E. Martin, and D. F. DeSante. 1993. Handbook of field methods for monitoring landbirds. U.S. Department of Agriculture, Forest Service, Pacific Southwest Research Station, Albany, California, USA. https://doi.org/10.2737/PSW-GTR-144
R Core Team. 2020. R: A Language and Environment for Statistical Computing, Version 3.6.3. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/
Riva, F., J. H. Acorn, and S. E. Nielsen. 2018a. Narrow anthropogenic corridors direct the movement of a generalist boreal butterfly. Biology Letters 14:20170770. https://doi.org/10.1098/rsbl.2017.0770
Riva, F., J. H. Acorn, and S. E. Nielsen. 2018b. Localized disturbances from oil sands developments increase butterfly diversity and abundance in Alberta’s boreal forests. Biological Conservation 217:173-180. https://doi.org/10.1016/j.biocon.2017.10.022
Riva, F., and S. E. Nielsen. 2020. A functional perspective on the analysis of land use and land cover data in ecology. Ambio 50:1089-1100. https://doi.org/10.1007/s13280-020-01434-5
Robin, X., N. Turck, A. Hainard, N. Tiberti, F. Lisacek, J.-C. Sanchez, and M. Müller. 2011. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 12:77. https://doi.org/10.1186/1471-2105-12-77
Rosa, L., K. F. Davis, M. C. Rulli, and P. D’Odorico. 2017. Environmental consequences of oil production from oil sands. Earth’s Future 5:158-170. https://doi.org/10.1002/2016EF000484
Schieck, J., and S. J. Song. 2006. Changes in bird communities throughout succession following fire and harvest in boreal forests of western North America: literature review and meta-analyses. Canadian Journal of Forest Research 36:1299-1318. https://doi.org/10.1139/x06-017
Schmiegelow, F. K. A., C. S. Machtans, and S. J. Hannon. 1997. Are boreal birds resilient to forest fragmentation? An experimental study of short-term community responses. Ecology 78:1914. https://doi.org/10.1890/0012-9658(1997)078[1914:ABBRTF]2.0.CO;2
Schneider, R. R., and S. Dyer. 2006. Death by a thousand cuts: impacts of in situ oil sands development on Alberta’s boreal forest. Canadian Parks and Wilderness Society, Edmonton, Alberta, Canada, and Pembina Institute, Calgary, Alberta, Canada.
Schneider, R. R., J. B. Stelfox, S. Boutin, and S. Wasel. 2003. Managing the cumulative impacts of land uses in the Western Canadian Sedimentary Basin: a modeling approach. Conservation Ecology 7:8. https://doi.org/10.5751/ES-00486-070108
Scrafford, M. A., T. Avgar, B. Abercrombie, J. Tigner, and M. S. Boyce. 2017. Wolverine habitat selection in response to anthropogenic disturbance in the western Canadian boreal forest. Forest Ecology and Management 395:27-36. https://doi.org/10.1016/j.foreco.2017.03.029
Smith, A. C., L. Fahrig, and C. M. Francis. 2011. Landscape size affects the relative importance of habitat amount, habitat fragmentation, and matrix quality on forest birds. Ecography 34:103-113. https://doi.org/10.1111/j.1600-0587.2010.06201.x
Sólymos, P., S. M. Matsuoka, E. M. Bayne, S. R. Lele, P. Fontaine, S. G. Cumming, D. Stralberg, F. K. A. Schmiegelow, and S. J. Song. 2013. Calibrating indices of avian density from non-standardized survey data: making the most of a messy situation. Methods in Ecology and Evolution 4:1047-1058. https://doi.org/10.1111/2041-210X.12106
Sólymos, P., S. M. Matsuoka, D. Stralberg, N. K. S. Barker, and E. M. Bayne. 2018. Phylogeny and species traits predict bird detectability. Ecography 41:1595-1603. https://doi.org/10.1111/ecog.03415
Stern, E. R., F. Riva, and S. E. Nielsen. 2018. Effects of narrow linear disturbances on light and wind patterns in fragmented boreal forests in northeastern Alberta. Forests 9:1-13. https://doi.org/10.3390/f9080486
St-Louis, V., A. M. Pidgeon, T. Kuemmerle, R. Sonnenschein, V. C. Radeloff, M. K. Clayton, B. A. Locke, D. Bash, and P. Hostert. 2014. Modelling avian biodiversity using raw, unclassified satellite imagery. Philosophical Transactions of the Royal Society B: Biological Sciences 369:20130197. https://doi.org/10.1098/rstb.2013.0197
Thomas, E. H., M. C. Brittingham, and S. H. Stoleson. 2014. Conventional oil and gas development alters forest songbird communities. Journal of Wildlife Management 78:293-306. https://doi.org/10.1002/jwmg.662
Tigner, J., E. M. Bayne, and S. Boutin. 2014. Black bear use of seismic lines in Northern Canada. Journal of Wildlife Management 78:282-292. https://doi.org/10.1002/jwmg.664
Tigner, J., E. M. Bayne, and S. Boutin. 2015. American marten respond to seismic lines in northern Canada at two spatial scales. PLoS ONE 10:1-19. https://doi.org/10.1371/journal.pone.0118720
Turgeon, P. J., S. L. Van Wilgenburg, and K. L. Drake. 2017. Microphone variability and degradation: implications for monitoring programs employing autonomous recording units. Avian Conservation and Ecology 12:9. https://doi.org/10.5751/ACE-00958-120109
Van Rensen, C. K. 2014. Predicting patterns of regeneration on seismic lines to inform restoration planning in boreal forest habitats. Thesis. University of Alberta, Edmonton, Alberta, Canada.
Van Rensen, C. K., S. E. Nielsen, B. White, T. Vinge, and V. J. Lieffers. 2015. Natural regeneration of forest vegetation on legacy seismic lines in boreal habitats in Alberta’s oil sands region. Biological Conservation 184:127-135. https://doi.org/10.1016/j.biocon.2015.01.020
Villard, M. A., and J. P. Metzger. 2014. Beyond the fragmentation debate: a conceptual model to predict when habitat configuration really matters. Journal of Applied Ecology 51:309-318. https://doi.org/10.1111/1365-2664.12190
Willier, C. 2017. Changes in peatland plant community composition and stand structure due to road induced flooding and desiccation. Thesis. University of Alberta, Edmonton, Alberta, Canada.
Yip, D. A., E. M. Bayne, P. Sólymos, J. Campbell, and D. Proppe. 2017. Sound attenuation in forest and roadside environments: implications for avian point-count surveys. Condor 119:73-84. https://doi.org/10.1650/CONDOR-16-93.1
Zuur, A. F., E. N. Ieno, and C. S. Elphick. 2010. A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution 1:3-14. https://doi.org/10.1111/j.2041-210X.2009.00001.x
Table 1. Summary of mean percent area (± SD) for habitat and development feature types within 100 m and 500 m of point-count centers for 157 stations used in occurrence models for Dark-eyed Juncos (Junco hyemalis) and Palm Warblers (Setophaga palmarum).
|Local-scale (100-m radius)||Neighborhood (500-m radius)|
|Variables||Mean percent area||Range||Mean percent area||Range|
|Lowland†||68.1 ± 19.1||20.7–100†||63.8 ± 16.7||19.7–96.9|
|Shrubby lowland||8.9 ± 19.2||0–100||8.7 ± 11.0||0–69.8|
|Treed bog||47.3 ± 31.8||0–100||44.7 ± 24.5||0–95.2|
|Other habitats||5.9 ± 12.9||0–69.2||11.3 ± 14.0||0–63.6|
|Polygonal permanent||2.9 ± 11.4||0–62.3||3.8 ± 9.2||0–57.3|
|Linear permanent||2.7 ± 8.3||0–42.4||2.7 ± 3.7||0–12.6|
|Well sites||2.1 ± 5.3||0–24.1||1.3 ± 1.0||0–5.6|
|Wide linear||4.2 ± 6.9||0–44.3||3.7 ± 2.9||0–13.3|
|Seismic||13.9 ± 10.2||0–30.6||13.5 ± 9.2||0–25.4|
|†Stations selected for analysis included a minimum of 20% lowland habitat within 100 m of point-count center.|
Table 2. Akaike’s information criterion corrected for small sample sizes (AICc), changes in AICc (ΔAICc) and relative weights (wi) for predictive models of Dark-eyed Junco (Junco hyemalis) occurrence, where K represents the number of model parameters and LL is the log-likelihood for the model.
|Habitat† + permanent polygonal500 + permanent linear10||4||-102.26||212.78||0||0.64|
|Habitat† + permanent linear50||3||-104.78||215.71||2.93||0.15|
|Habitat† + permanent polygonal500||3||-105.36||216.87||4.09||0.08|
|Habitat† + permanent polygonal500 + permanent linear100 + well site500 + wide linear100 + seismic500||7||-101.95||218.64||5.87||0.03|
|Habitat† + wide linear100||3||-106.93||220.01||7.23||0.02|
|Habitat† + seismic500||3||-107.11||220.38||7.6||0.01|
|Habitat† + well site500||3||-107.17||220.5||7.72||0.01|
|Habitat† + well site500 + wide linear100||4||-106.92||222.11||9.33||0.01|
|Habitat† + well site500 + wide linear100 + seismic500||5||-106.82||224.04||11.26||0|
|†Habitat includes the variables identified for the best supported “habitat only” null model. For Dark-eyed Juncos this includes treed bog within 500 m (Table 3, Appendix 1).|
‡Model variables were examined for collinearity and all models contained variance inflation factor (VIF) values < 3. For the Dark-eyed Junco “All features” model, the VIF values were 1.04- treed bog500, 1.12- permanent polygonal500, 1.07- permanent linear100, 1.24- well site500, 1.07-wide linear100, and 1.22- seismic500.
Table 3. Summary of parameter estimates (β), their associated standard error (SE) and probability, odds ratios, 95% confidence intervals for the odds ratios, and standardized beta coefficients (βSTD) in the most supported Dark-eyed Junco (Junco hyemalis) occurrence model.
|Treed bog (500 m)||0.352||0.171||0.04||1.421||(1.035, 2.031)||0.388|
|Polygonal permanent (500 m)||-0.462||0.235||0.05||0.63||(0.368, 0.948)||-0.426|
|Linear permanent (100 m)||0.552||0.25||0.027||1.737||(1.116, 3.064)||0.460|
Table 4. Akaike’s information criterion corrected for small sample sizes (AICc), changes in AICc (ΔAICc) and relative weights (wi) for predictive models of Palm Warbler (Setophaga palmarum) occurrence, where K represents the number of model parameters and LL is the log-likelihood for the model.
|Habitat† + permanent polygonal100 + permanent linear500||6||-86.52||185.59||0||0.63|
|Habitat† + permanent linear500||5||-88.83||188.06||2.46||0.18|
|Habitat† + permanent polygonal100||5||-89.68||189.75||4.16||0.08|
|Habitat† + permanent polygonal100 + permanent linear500 + well site100 + wide linear500 + seismic500||9||-85.35||189.93||4.34||0.07|
|Habitat† + seismic500||5||-91.36||193.13||7.53||0.01|
|Habitat† + well site100||5||-91.84||194.08||8.49||0.01|
|Habitat† + well site100 + wide linear500||5||-92.06||194.52||8.93||0.01|
|Habitat + wide linear500||5||-92.06||194.52||8.93||0.01|
|Habitat† + well site100 + wide linear500 + seismic500||7||-90.78||196.3||10.71||0|
|Shrubby lowland500 + vegetation variability100 + vegetation productivity100||2||-101.07||206.22||20.63||0|
|†Habitat includes the variables identified for the best supported “habitat only” null model. For Palm Warblers this includes shrubby lowland within 500 m, vegetation productivity within 100 m, and vegetation variability within 100 m (Table 5, Appendix 1).|
‡Model variables were examined for collinearity and all models contained variance inflation factor (VIF) values < 3. For the Palm Warbler “All features” model the VIF values were 1.74- shrubby lowland500, 1.53- vegetation variability100, 2.05- vegetation productivity100, 1.03- permanent polygonal100, 1.29- permanent linear500, 1.14- well site100, 1.09- wide linear500, and 1.22- seismic500.
Table 5. Summary of parameter estimates (β), their associated standard error (SE) and probability, odds ratios, 95% confidence intervals for the odds ratios, and standardized beta coefficients (βSTD) in the best supported Palm Warbler (Setophaga palmarum) occurrence model.
|Shrubby lowland (500 m)||0.219||0.102||0.031||1.245||(1.025, 1.532)||0.535|
|Vegetation variability (100 m)||-0.993||0.272||0||0.37||(0.211, 0.617)||-0.867|
|Vegetation productivity (100 m)||0.244||0.092||0.008||1.277||(1.076, 1.545)||0.688|
|Polygonal permanent (100 m)||-0.813||0.648||0.209||0.443||(0.052, 0.954)||-0.928|
|Linear permanent (500 m)||1.37||0.562||0.015||3.936||(1.346, 12.348)||0.506|