Research Paper

INTRODUCTION

Songbirds in the Parulidae family represent one of the groups with the highest rates of decline in North America (Rosenberg et al. 2019). In particular, long-distance migratory species associated with old-growth forests experience major threats (Lampila et al. 2005). The Canada Warbler (Cardellina canadensis) is a migratory songbird commonly associated with old-growth forest in the boreal biome and has been declared a species at risk in Canada (Bayne et al. 2016, Environment Canada 2016) due to a substantial decline in abundance over the last half century (Sauer et al. 2020, Wells et al. 2020). Causes of its decline are thought to include habitat loss, habitat alteration, and changes to forest successional patterns on the breeding grounds (Government of Alberta 2010, Reitsma et al. 2010). Short harvest rotations and loss of older forest age classes may also be contributing to the decline (Grinde and Niemi 2016). Studies across the species’ range show mixed responses to forest harvesting and associated disturbance (Hunt et al. 2017). In Alberta, Canada Warbler has higher density and productivity in postharvest areas (harvested areas with regeneration > five years) than in recent clear-cuts (Ball et al. 2016, Hunt et al. 2017). There is also evidence of the detrimental effects of road networks on the density of some populations of Canada Warbler (Miller 1999, Haché et al. 2014, Westwood et al. 2019a).

Knowledge of the ecology and geographic distribution of a species are critical for prioritizing and informing conservation action, planning, and assessing threats from various anthropogenic factors (Akçakaya and Atwood 1997, Wintle et al. 2005, Hirzel et al. 2006). However, the current understanding of the ecological needs of Canada Warbler is biased by the selection of study sites (and associated findings), which are influenced by site accessibility (Environment Canada 2016). For example, projects such as the provincial Breeding Bird Atlases (BBAs) and the Breeding Bird Survey (BBS) are restricted mainly to roadside surveys (Kirk et al. 1997, Matsuoka et al. 2011), whereas most of the Canada Warbler range is in areas with low road density or no roads at all. Additionally, the species’ apparent habitat needs might be distorted by limitations in current habitat availability compared to historical conditions (Environment Canada 2016, Wells et al. 2018). Species distribution models (SDMs) can be a useful tool for understanding habitat associations and identifying conservation and management opportunities, particularly in under sampled areas. These models are usually correlative (Guisan and Zimmermann 2000) and quantify the relationship between field observations and a set of environmental variables that are expected to reflect some key aspects of the species–habitat association (Hirzel et al. 2006). The resulting spatial predictions of species distribution, and associated maps of these estimates, are widely used to guide conservation strategies (McShea 2014).

For Canada Warbler, five SDMs have already been developed; two at a national scale in Canada (Haché et al. 2014, Stralberg et al. 2015) and another three at regional scales in Alberta (Ball et al. 2016) and the Atlantic provinces; in the latter case, one at 150 m x 150 m resolution (Bale et al. 2020) and another at 250 m x 250 m resolution (Westwood et al. 2019a). Canada Warbler is known to exhibit regional variation in habitat associations, particularly between the eastern and western portions of its range (Leston et al. 2024), and regionally specific habitat associations are needed to guide local management priorities. In Northwestern Ontario, little information about Canada Warbler is available. Quetico Provincial Park (Fig. 1), one of the largest wilderness-protected areas in Northwestern Ontario (covering 2,035,903 ha that represents 2.0% of the province; Crins et al. 2009), remains heavily forested compared to other regions. Currently, most of Northwestern Ontario either has been logged or is projected to be logged in the near future, leaving Quetico the most extensive area without timber extraction. The southern portion of Northwestern Ontario, encompassing the Quetico area, has one of the highest relative abundances of the Canada Warbler in the province, with counts above three–ten birds/route/year according to the BBS project (Sauer et al. 2020; Appendix 1, Fig. A1.1).

Most SDMs are only validated statistically by using part of the original dataset applied as a test dataset (Roberts et al. 2017, Westwood et al. 2019b). The best way to validate the accuracy of an SDM is using independent, field-collected data (Yates et al. 2018); however, this step is rarely taken because of a lack of time or funding to collect and process new data (Franklin 2013). Also, notwithstanding thousands of published SDMs in the past two decades, relatively few examples are validated by an independently collected dataset (e.g., Ortega-Huerta and Vega-Rivera 2017, Westwood et al. 2019b). We built two SDMs using maximum entropy algorithm, Maxent (Phillips et al. 2006), and we field-validated the model by collecting independent field observations in 2022. Our objectives were to (1) predict the distribution of Canada Warbler in an ecoregion of Northwestern Ontario; (2) compare habitat associations and probability of occurrence between protected and logged areas to assess how habitat alteration influences distribution; and (3) use an independent field-collected and ground-truthed dataset to validate the accuracy of our final model and measure correspondence between predicted and observed occurrences. The final model can be used to guide conservation and management action for this species at risk and inform methods for modeling Canada Warbler or other migratory bird species in other regions.

METHODS

Study area

Northwestern Ontario encompasses eight ecoregions (0E, 1E, 2W, 3S, 4S, 5S, 3W, and 4W; Crins et al. 2009); BBS and OBBA have collected location data for Canada Warbler mainly restricted to the southern portion of the region due to the limited accessibility. We accordingly restricted our study area to the 4W Ecoregion (also known as the Pigeon River Ecoregion), where the Canada Warbler has a higher relative abundance (Sauer et al. 2020, Appendix 1, Fig. A1.1) and where Quetico Provincial Park is located (Crins et al. 2009, Fig. 1). The climate of this ecoregion is cool and relatively dry; the mean annual temperature is -0.2–2.7 °C, and the mean growing season length is 168–188 days (Ontario Climatic Model v.2.0 [CD-ROM], Ontario Ministry of Natural Resources 2000, V. Cupiche-Herrera, personal observation). Annual precipitation ranges from 674–838 mm, and mean summer rainfall ranges from 225–300 mm (Crins et al. 2009). Mixed forest is the most extensive land cover class (33.2%), sparse forest occurs at 19.3%, water at 17.5%, coniferous forest at 11.5%, deciduous forest at 10.6%, and cutovers at 3.6% of the area of this ecoregion.

Ecoregion 4W is divided into two ecodistricts, 4W-1, Quetico and 4W-2, Kakabeka. Ecodistrict 4W-1 has human settlement on <1% of its area: the largest community is Atikokan, Ontario, and the protected areas encompass 30%, including Quetico Provincial Park (4,760 km²), the first official park in Northwestern Ontario (Wester et al. 2018). Ecodistrict 4W-2, Kakabeka, represents 18.1% of the ecoregion, and settlement and associated infrastructure cover 2% of the total area, whereas protected areas cover only 4.3% of the ecodistrict. Predominant land uses include timber harvesting, resource-based tourism, mineral exploration, and agriculture; the city of Thunder Bay (Fig. 1) is the largest urbanized community in Northwestern Ontario. According to the last census in 2021 (Statistics Canada 2022), Thunder Bay district has a population of 146,867, with 74.1% living in the city of Thunder Bay (population 108,849).

Species occurrence dataset

For our preliminary (2020) model, we used existing data sources of Canada Warbler occurrences. We used occurrence data derived from the long-term songbird monitoring program in Quetico Provincial Park (2014–2019), BBS (Ziolkowski et al. 2022), Ontario Breeding Bird Atlas (OBBA; Bird Studies Canada, 2001–2005), and the Cornell Lab of Ornithology’s eBird project (Fink et al. 2021), which has compiled avian point count data in North America from incident observation records collected through community science programs. The eBird project can offer an extensive coverage dataset for understudied or elusive species at no cost to the user (Sullivan et al. 2014). Community science datasets like eBird can provide important information outside of regularly surveyed areas (Lin et al. 2022) and support design and management of protected areas (Binley et al. 2021). The eBird dataset is recorded by a variety of people: amateurs, tourists, researchers, and volunteers, sometimes collecting data farther than from roadside.

From the four databases (eBird, OBBA, BBS, and Quetico Provincial Park), we extracted 235 Canada Warbler presence observations. We applied a spatial filtering of 250 m to the occurrence dataset; we used this distance because the general recommendation is thinning the data by the typical home range of the species to avoid overlapping and repetition of individuals (Boria et al. 2014, Aiello-Lammens et al. 2015). The home-range sizes of the species ranges from 1–3.3 ha, and there is evidence that individual home ranges overlap (Environment Canada 2016, Flockhart et al. 2016, Hunt et al. 2017). When the distance between two points was less than 250 m, the point closer to a road was removed (Bale et al. 2020). We removed 66 such duplicated observations, and we removed localities with missing coordinates and other georeferencing errors as well as data falling over areas covered by clouds in the Landsat layers (Boria et al. 2014, Cobos et al. 2018). We excluded 93 occurrences in total, leaving 78 observations (from 2001–2020) for the construction of a 2020 preliminary model. Given that 78 observations is a relatively small dataset, especially for an area of this size, we used the model 2020 to guide additional field surveys in spring of 2021. During that season, we surveyed a total of 132 point counts, obtaining 44 new observations of Canada Warbler to add to our training dataset. We ended with a total of 122 observations available for a second model (2021) and the final model (2022), with a timespan from 2001–2021 (Fig. 1; Appendix 1, Table A1.1).

Our field surveys followed the OBBA survey protocol, and we recruited volunteers to assist with counts during the breeding seasons (late May to late July) of 2021 (additional data for training dataset for the final model) and 2022 (independent data for model testing). OBBA uses point counts within a 10 x 10 km² area, recording birds for five minutes at each point approximately 250–300 m apart from each other. For 2022, we surveyed a total of 118 locations and obtained 30 new Canada Warbler observations to use as validation data for the final model. We recorded the coordinates of each point count location using a GPS unit. The two observers conducted the point counts in favorable weather conditions, on days without precipitation or wind > 20 km/h, both of which reduce the detectability of singing birds (Cadman et al. 2007).

After the 5-min bird point count, observers used speakers or mobile devices to play Canada Warbler songs to increase the detectability of the species and register the number of individuals in the area. The observers recorded the total number of territorial males detected by sight or sound at each point count using the following protocol: (1) 30 s of playbacks of conspecific songs, (2) 1 min silence, (3) 30 s of playbacks, and (4) 1 min silence (protocol modified from Flockhart et al. 2016, Hunt et al. 2017). The repetition of the playback and the silent periods helped to reduce the bias of artificial calls and the possible effect of individuals approaching just because of curiosity. The observers counted only the males that responded during or after the second period of playback of conspecific songs.

Environmental covariates

We selected variables for the SDM that most closely matched the Canada Warbler niche as described by the literature. The species inhabits many forest types but is most abundant in humid, mixed-wood forests with a dense understory and complex ground cover; it is associated with forest disturbance that creates suitable understory conditions (Becker et al. 2012). Local-scale studies have suggested that the Canada Warbler has breeding territories often on steep slopes near streams (Schieck et al. 1995, Schieck and Song 2006, Reitsma et al. 2010). On larger scales (landscape and regional), Haché et al. (2014) found Canada Warbler densities are generally higher in areas with tall trees, and Westwood et al. (2019a) found the species to be sensitive to anthropogenic disturbance. Based on this information, we initially selected ten variables that captured local and landscape features (Table 1).

We created geospatial raster layers in QGIS (QGIS.org 2021, version 3.18) to extract the candidate variables as predictors or covariates. To get a high resolution SDM, we used Landsat 30 m resolution images. However, because of the climate conditions of the region (high humidity and condensation) we could not find cloud-free (< 10%) images from the whole Ecoregion to extract candidate predictor variables from different years. We were limited to using cloud-free (< 10%) satellite images from Landsat 8 Collection 2 Level–1 from June and August 2018 courtesy of the U.S. Geological Survey (2020).

We calculated spectral indices from Landsat images to use as covariates. Spectral indices included calculating bare soil index (BASI) as an indicator of soil uncover and open areas such as grasslands, shrublands, clear-cuts, and roads. We extracted the enhanced vegetation index (EVI) and normalized difference vegetation index (NDVI) as indicators of vegetation cover. Land surface temperature (LST) is an indicator of environmental temperature. The normalized water index (NDWI) was extracted because it is associated with changes in vegetation water content and water absorption from tree canopies, serving as an indicator of deciduous vegetation (trees and shrubs). The normalized difference moisture index (NDMI) is associated with vegetation moisture and indicators of wetlands. We obtained the digital elevation model (DEM) to reflect the elevation of the terrain, and we calculated the distance to water bodies (WATER) based on the known association of Canada Warbler with riparian zones.

Another set of covariates was calculated from other data sources besides Landsat. We calculated the distance to mature coniferous forests (D_CONIF) using the Ontario Land Cover Compilation produced by the Ontario Ministry of Natural Resources, publicly available through Ontario GeoHub (https://geohub.lio.gov.on.ca/). In addition, we obtained spatial layers with information on global forest canopy height (CAN; Potapov et al. 2021). To identify when stands changed from a forest to non-forest state (usually through logging but occasionally through wildlife; DISTURB), we used the Global Land Analysis and Discovery dataset from the University of Maryland (Hansen et al. 2013; Potapov et al. 2021). All covariates were clipped using the Ecoregion 4W shape file freely available through the Ontario GeoHub webpage. All datasets, their resolution, and source information are given in Table 1.

Model construction

We selected the algorithm Maximum Entropy (Phillips et al. 2021; Maxent software version 3.4.1). Maxent estimates a target probability distribution by means of knowing the maximum entropy distribution and has the ability to handle observations collected using different protocols (Phillips et al. 2006). Maxent performs well compared to other modeling methods (Elith et al. 2006), including when small dataset presence data are available, and does not require absence data (Hernández et al. 2006), making it useful in data-poor regions (Phillips et al. 2006).

We ran a preliminary model (2020), where we used the desktop-collected dataset from 2001–2020 (78 observations) and cross-validation with a random seed subset (25%) of the data. We ran a second model (2021) using desktop-collection dataset from 2001–2021 and the 2021 field observations (for a total of 122 Canada Warbler observations) and cross-validation with a random seed subset (25%) of the data. For the final model (2022), we the full dataset from 2001–2021 (N= 122), and we cross-validated with the field-collected data from 2022 (30 observations).

To reduce the collinearity among the environmental covariates, we used QGIS GRASS correlation analysis (QGIS.org 2021, version 3.18) to select the pairs of variables with Pearson correlation coefficient values r ≥ 0.7, and, in those cases where sets of covariates were correlated, we used the covariate that better matched the geographic distribution of the species in preliminary model testing (Dormann et al. 2012). We tested all covariates for independence using a correlation matrix, which assesses relationships between variables (Appendix 1, Table A1.2). Highly correlated covariates were NDMI with BASI and EVI, and NDVI with BASI and NDMI.

We removed NDMI and NDVI from the first steps of model construction. We initially ran the models with all the remaining covariates, later identifying LST as a covariate that did not contribute to any of the models and reduced the model performance, so we removed LST from the final modeling steps. For the preliminary model (2020) we also removed BASI because it did not contribute to the model and reduced its performance.

For all models we treated the variables as continuous, and we ran 10 cross-validation replicates. We used a jackknife approach to test model fit and assess the model’s prediction, expressed as the area under the curve (AUC) for both training and test data. The standard statistical method to assess model accuracy is derived from the Receiver Operating Characteristic (ROC) curves, created by plotting sensitivity (proportion of observed occurrences correctly predicted) against specificity (proportion of observed absences correctly predicted) for all possible thresholds (Pearson 2010). With the presence-only methods like Maxent, specificity cannot be calculated so the standard is to use pseudo-absences (random background points) instead of absences (Lahoz-Monfort et al. 2010). An AUC value of 0.5 shows that model predictions are not better than random, < 0.5 are worse than random, 0.5–0.69 indicates poor performance, 0.7–0.9 reasonable/moderate performance, and > 0.9 indicates high performance (Peterson et al. 2011).

Including areas where the species cannot occur or for which there are low presence data (such as the less accessible ecoregions in the Northwestern Ontario range of Canada Warbler) will generally predict lower suitability levels and create a model that appears to have high performance (Barve et al. 2011). In other words, increasing the extent of the prediction area can include absences or pseudo-absences that inflate AIC values by parameterizing models with good discrimination capacity that are low in information (Jiménez-Valverde et al. 2008). Therefore, we prioritized a more accurate model over one with high performance but lacking objective evaluation.

Field-validation

To field-validate the model accuracy in Ecoregion 4W, we developed a binary prediction from the initial SDM of 2021 (Appendix 1, Fig. A1.2). We divided the prediction into two classes of Maxent probability of occurrence due to restricted accessibility in many areas and for maximizing the detection probabilities for Canada Warbler: high predicted occurrence (index > 0.60) and low predicted probability of occurrence (≤ 0.59). We collected data during the breeding season of 2022 for both prediction classes, by randomly selecting locations within 100 m of main roads and with a minimum separation of 250 m among point counts. We identified for sampling 46 sites with high predicted occurrence and 72 sites with low predicted occurrence. Model prediction accuracy was classified using an error matrix by comparing the prediction of the 2021 model with the reference data collected on the ground in 2022. We used Cohen’s Kappa (Cohen 1960) coefficient (K) to compare the correspondence between predicted and observed occurrences.

(1)

Pa is the relative observed agreement (level of agreement among each category divided by the total number of observations) and Pe is the expected agreement by chance. Values below 0.4 are considered poor agreement, between 0.4 and 0.5 is the median agreement level, and above 0.6 is considered a high accuracy.

RESULTS

Model performance

Across 10 replicates, the preliminary model (2020) of probability of occurrence of Canada Warbler showed poor performance for both training data (AUC_training average 0.69) and test data (AUC_test average 0.64). The second model (2021) showed moderate performance for training data (AUC_training average 0.75) and poor to moderate performance for test data (AUC_test average 0.71). The final model (2022) had similar performance to the second model (2021) but better than the preliminary model (Fig. 2), with moderate performance for training data (AUC_training average 0.75) and poor to moderate for test data (AUC_test average 0.68).

For the preliminary model (2020), we retained seven covariates (Table 2). The covariates with high mean percent contribution (> 11%) included WATER (distance to water bodies), DISTURB (years since disturbance, mainly due to logging), and EVI (enhanced vegetation index). The covariates that had a lower percent of contribution (< 11%) were NDWI (normalized difference water index), DEM (digital elevation model), D_CONIF (distance to conifer forest), and CAN (tree canopy height). For the second model (2021), we retained eight covariates (Table 2). The covariates with high mean percent contribution (> 11%) included NDWI, WATER, EVI, D_CONIF. The covariates that have a lower percent of contribution (< 11%) were DEM, CAN, DISTURB, and BASI (bared soil index).

The final model (2022) used the same covariates as the 2021 model (Table 2). The covariates that contributed more to the model (> 11%) where NDWI, WATER, EVI, and D_CONIF. The covariate that contributed most to the final model was NDWI, where the prediction of Canada Warbler decreased while higher vegetation water content increased (Fig. 3A). The second covariate was WATER, the Canada Warbler probability of occurrence increased closer to the water bodies (Fig. 3B). EVI was the covariate that contributed third most to the model, and, as EVI increased, the probability of Canada Warbler was predicted to be higher up to an EVI = 0.6 and lower beyond this value (Fig. 3C), indicating a high association with shrubs and medium-density forested areas. The covariate that contributed fourth most to the model was D_CONIF, whereby probability of Canada Warbler occurrence decreased with increasing distance to the coniferous forest (Fig. 3D).

Covariates with a lower mean percentage contribution to the model (≤ 11%, Table 2) were CAN (tree canopy height), DEM (the digital elevation model), DISTURB (years since disturbance by logging), and BASI (bare soil index). The response curve of CAN (Fig. 4A) indicates Canada Warbler probability of occurrence increased with the canopy tree height. According to DEM, the highest predicted probability of occurrence of the Canada Warbler (above 0.8) is at 400 m elevation (Fig. 4B). The response curve of DISTURB indicates Canada Warbler had a higher predicted probability of occurrence in areas where no disturbance happened during the last 20 years and a lower predicted probability of occurrence when disturbance was detected more recently (Fig. 4C); observations of Canada Warbler in areas of forest disturbance are reported in Appendix 1, Table A1.3. Finally, the covariate that contributed least to the model was BASI; the association to low values (< 0) of BASI indicates association to high shrub density (Fig. 4D).

Field-validation results

From the total 118 sites surveyed in 2022, we identified 88 matches for both categories (low and high predicted areas), achieving 74.57% total accuracy for the SDM. In areas where the preliminary model showed high probability of occurrence of the Canada Warbler (index > 0.60), we observed an accuracy of 50%, and identified Canada Warbler males in 23 of the 46 surveyed total sites. The accuracy was 90.27% in sites with a low prediction of occurrence; we observed Canada Warbler males only in seven sites (9.72%) from a total of 72 sites with low model prediction. The two observed class conditions received a Kappa value of 0.43, indicating agreement better than expected by chance.

DISCUSSION

This is the first study to develop a high-resolution field-validated SDM for Canada Warbler in Northwestern Ontario. The AUC value (0.68–0.75) is acceptable, and the field-validation process also indicates that our model performed better than expected by chance. The AUC value in our model is similar to the Maxent model developed by Bale et al. (2020) for Nova Scotia (AUC = 0.7) with a larger observations’ dataset. Even though our dataset has a limited number of observations (N = 122), the moderate performance may be due to the inclusion of systematic field-collected observations in our training dataset, the validation through the ground-truth process, and the use of more current Landsat images to develop the indexes used as covariates. The majority of Canada Warbler models are done solely by large datasets of observations collected mainly by volunteers, who include highly knowledgeable birders but also novices who can misidentify, overestimate, or under-detect the species in the field. There is a high chance of species under-detection by the observers who registered their observations in eBird, OBBA, and BBS, a reason we did not try to infer absences records from these datasets. The poor performance (AUC < 0.7) of the preliminary model (2020) is because of its construction mainly with those datasets.

Previous studies on Canada Warbler’s habitat in the eastern portion of its range have identified vegetated wetlands and moist forests as important habitat types for this species (e.g., Goodnow and Reitsma 2011, Westwood et al. 2019a). We observed a high prediction of occurrence closer to water bodies (WATER), which may indicate that Canada Warbler is more likely to occur in proximity to riparian and wet areas. The association with the NDWI was mostly negative, indicating a strong relationship with deciduous trees and shrubs. Also, the response to bare soil (BASI) and the association with EVI suggests that relative open forested areas and denser understory are important predictors for Canada Warbler occurrence, similar to what Bale et al. (2020) found and confirming previous findings (e.g., Becker et al. 2012). Bale et al. (2020) also found that Canada Warbler is associated with a relatively close distance coniferous stands, being consistent with our model prediction that the occurrence of the species is higher in areas closest to coniferous forests. This finding may reflect a genuine association with habitats near to coniferous forests or may reflect the patchy nature of deciduous and/or mixed-wood forests that occur in a heterogenous matrix among coniferous forest.

Overall, our results for canopy height (CAN) were consistent with those observed by Haché et al. (2014), where Canada Warbler densities were generally higher in areas with tall trees (a proxy for forest age). Though canopy height was not one of the strongest predictor variables, taking this in combination with associations with mixed-wood forest, we caution that Canada Warbler may be vulnerable to short-rotation, even-aged forest management, which reduces the availability of old forests on the landscape and promotes the conversion of mixed-wood and coniferous-dominated forest landscapes to forests with broadleaf dominance (Drapeau et al. 2000, Hobson and Bayne 2000, Imbeau et al. 2001, Schieck and Song 2006).

Our final model predicted high probability of occurrence of Canada Warbler (> 0.7) in areas with no forest disturbance (Fig. 4C). These results are consistent with patterns observed in a study in Northern Minnesota, which found that Canada Warbler is more common in wilderness forest than in managed forest (Zlonis and Niemi 2014). However, there are also high probabilities of occurrence (> 0.6) in areas where the forest has been recently managed for timber extraction or post-fire disturbance (within 6 years since disturbance). Other studies reported a high density of Canada Warbler in postharvest cuts (regenerated areas > 5 years since disturbance; Ball et al. 2016, Hunt et al. 2017) and light partial harvest (Becker et al. 2012), indicating that the species can take advantage of logged regenerated areas with high shrub density. This difference may reflect regional variation in habitat associations (Crosby et al. 2019; Leston et al. 2024) or differences in the types of harvesting and silvicultural techniques used in different jurisdictions.

Limitations in model application

Canada Warbler has been reported as likely influenced by conspecific attraction (Flockhart et al. 2016, Hunt et al. 2017), and our model prediction was not able to integrate the influence of this behavior. We could have created a covariate that served as a proxy for conspecific attraction such as distance to nearest other Canada Warbler observations. However, given the small number of occurrences across a large area (N = 122) and the strong level of spatial bias in some of the occurrence dataset, we felt the level of accuracy in such a layer was too low to warrant its inclusion. Furthermore, density alone can be misleading and not necessarily indicate habitat quality (Van Horne 1983). We encourage managers to perform a field validation of the habitat suitability prediction (e.g., Westwood et al. 2019b) for the species, because high congregation of individuals on postharvest areas and low abundance in protected areas could not necessarily reflect the habitat quality of the Canada Warbler.

Part of the aim of this study was to develop a high resolution SDM for when we need to use remote sensing data such as Landsat. This method also is particularly relevant in cases where other traditionally used environmental variables are difficult to collect or not relevant for the extent of the area of study (Lahoz-Monfort et al. 2010), such as bioclimatic variables that are commonly used at larger spatial extents (e.g., Stralberg et al. 2015, Cadieux et al. 2020). However, the use of remote sensing images for specific or a limited number of years may reduce the application of the model, because the predictions will be closely tied to the environment conditions at the time the images were obtained. An issue with Landsat is the persistent cloud cover in certain regions, more than other sensors of its 16–18-day revisit period (Wijedasa et al. 2012); solving this problem may be possible, but it is time consuming, and it requires specialized skills in GIS that most researchers and land managers often do not have. Despite its limitations, Landsat imagery has good qualities that help monitoring regional changes to forest habitat and biodiversity (Wijedasa et al. 2012).

One of our greatest limitations was that available occurrence data for this area is scarce, and with several inaccuracies (wrong coordinates, duplicated identifications, etc.), it is important to encourage volunteers and community naturalists to not only collect more observations, but also, and more importantly, to verify their observations and locations before they upload them to eBird, iNaturalist, or OBBA datasets. Also, more systematic bird survey research will be beneficial to increase sample size and bird population knowledge. Moreover, avian surveys using autonomous recording units (ARUs) can overcome major limitations experienced by point count methods, including site access limitations associated with remote locations as well as the disruption of surveys due to inclement weather (Drake et al. 2021). Automated recording units can provide long-term and systematic species monitoring data for elusive species such as Canada Warbler; Quetico Provincial Park is carrying out its bird surveys through the use of ARUs, but the limited equipment, staff, and accessibility still affect the number of sites at which ARUs are located. The use of ARUs in protocols such as OBBA, BBS, and even for forestry managers could increase species records and improve the accuracy of SDMs developed in the region.

Model accuracy and implications for management

The field-validation process demonstrated that the model accurately predicted areas delimited at medium–low occurrence (90.27%). Thus, when using model results to identify locations to conserve or manage habitat for Canada Warbler, users can reliably exclude all areas with a predicted probability of occurrence ≤ 0.59 from the search. Though the accuracy of the model when predicting high occurrence areas (≥ 0.60) was not exceptional, these areas can be identified for further ground truth validate. We reiterate past calls for developers of SDMs to field-ground-truth validate their models, because this step not only provides assurances of the accuracy of the model, but also can make it simpler for managers to implement conservation and management plans with more confidence.

The number of occurrences of Canada Warbler, particularly in Quetico Provincial Park, is low not only because of the inconspicuous characteristics of the species, but also to low accessibility in the park, where the main interests among visitors are canoeing and fishing, not recording bird observations. Quetico Provincial Park will benefit from increasing the interest of visitors in recording bird observations and uploading them into public datasets (e.g., iNaturalist, eBird). The present species distribution model could be used to guide both park staff and visitors to search for Canada Warbler within the provincial park in areas with higher probability of occurrence.

Critical habitat identification for species at risk in Canada is mandated (Government of Canada 2023). The results from our study could be important to land managers and logging companies in the ecoregion that need to consider the protection of species at risk in land use planning. At the site level, when conducting forestry or land-clearing activities during the breeding season, operators should search all areas with a probability of occurrence > 0.60 for Canada Warbler prior to incurring disturbance to mitigate potential impacts to this listed species at risk. In terms of landscape-level harvesting plans, it is necessary to preserve large, unharvested fragments or implement the retention of important forest features for the species during forest management, such as tall trees. Our model predicted Canada Warbler has mid-to-high probability of occurrence in various post-harvested areas, but the species is still relying on features associated with old-growth forest. In Ontario, 86% of harvested forests are under a clear-cut silviculture system (Ontario Ministry of Natural Resources and Forestry 2020); shifting to a more uneven-aged forest management system will be beneficial to Canada Warbler conservation. Also, through the SDM, it was detected that Canada Warbler seems to have high association with riparian zones, and increasing the buffering area of these zones from logging should be beneficial in the conservation and management of this species at risk.

LITERATURE CITED

Aiello-Lammens, M. E., R. A. Boria, A. Radosavljevic, B. Vilela, and R. P. Anderson. 2015. spThin: an R package for spatial thinning of species occurrence records for use in ecological niche models. Ecography 38: 541-545. http://doi.org/10.1111/ecog.01132

Akçakaya, H. R., and J. Atwood. 1997. A habitat-based metapopulation model of the California Gnatcatcher. Conservation Biology 11:422-434. https://doi.org/10.1046/j.1523-1739.1997.96164.x

Bale, S., K. F. Beazley, A. Westwood, and P. Bush. 2020. The benefits of using topographic features to predict climate-resilient habitat for migratory forest landbirds: an example for the Rusty Blackbird, Olive-sided Flycatcher, and Canada Warbler. Condor: Ornithological Applications 122:1-19. http://doi.org/10.1093/condor/duz057

Ball, J. R., P. Sólymos, F. K. A. Schmiegelow, S. Hache, J. Schieck, and E. Bayne. 2016. Regional habitat needs of a nationally listed species, Canada Warbler (Cardellina canadensis), in Alberta, Canada. Avian Conservation and Ecology 11(2):10. http://doi.org/10.5751/ACE-00916-110210

Barve, N., V. Barve, A. Jiménez-Valverde, A. Lira-Noriega, S. P. Maher, A. T. Peterson, J. Soberón, and F. Villalobos. 2011. The crucial role of the accessible area in ecological niche modeling and species distribution modeling. Ecological Modelling 222:1810-1819. http://doi.org/10.1016/j.ecolmodel.2011.02.011

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, F. K. A. Schmiegelow, and S. S. Song. 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

Becker D. A., P. B. Wood, and P. D. Keyser. 2012. Canada Warbler use of harvested stands following timber management in the southern portion of their range. Forest Ecology and Management 276:1-9. https://doi.org/10.1016/j.foreco.2012.03.018

Binley A. D., C. A. Proctor, R. Pitcher, S. A. Davis, and J. Bennett. 2021. The unrealized potential of community science to support research on the resilience of protected areas. Conservation Science and Practice 3:e376. https://doi.org/10.1111/csp2.376

Boria, R. A., L. E. Olson, S. M. Goodman, and R. P. Anderson. 2014. Spatial filtering to reduce sampling bias can improve the performance of ecological niche models. Ecological Modelling 275:73-77. https://doi.org/10.1016/j.ecolmodel.2013.12.012

Cadieux, P., Y. Boulanger, D. Cyr, A. R. Taylor, D. Price, P. Sólymos, D. Stralberg, H. Y. H. Chen, A. Brecka, and J. A. Tremblay. 2020. Projected effects of climate change on boreal bird community accentuated by anthropogenic disturbance in western boreal forest, Canada. Diversity and Distributions 26:668-682. https://doi.org/10.1111/ddi.13057

Cadman, M. D., D. A. Sutherland, G. G. Beck, D. Lepage, and A. R. Couturier. 2007. Atlas of the breeding birds of Ontario, 2001-2005. Bird Studies Canada, Environment Canada, Ontario Field Ornithologists, Ontario Ministry of Natural Resources, and Ontario Nature, Toronto, Ontario, Canada. https://www.birdsontario.org/

Cobos, M. E., L. Jiménez, C. Nuñez-Penichet, D. Romero-Alvarez, and M. Simoes. 2018. Sample data and training modules for cleaning biodiversity information. Biodiversity Informatics 13:49-50. https://doi.org/10.17161/bi.v13i0.7600

Cohen, J. 1960. A coefficient of agreement for nominal scales. Educational and Psychological Measurement 20:37-46.

Crins, W. J., P. A. Gray, P. W. C. Uhlig, and M. C. Wester. 2009. The ecosystems of Ontario, part I: ecozones and ecoregions. Inventory, Monitoring and Assessment Section, Science and Information Branch Technical Report TR- 01, Ontario Ministry of Natural Resources, Peterborough, Ontario, Canada. https://files.ontario.ca/mnrf-ecosystemspart1-accessible-july2018-en-2020-01-16.pdf

Crosby, A. D., E. M. Bayne, S. G. Cumming, F. K. A. Schmiegelow, F. V. Dénes, and J. A. Tremblay. 2019. Differential habitat selection in boreal songbirds influences estimates of population size and distribution. Diversity and Distributions 25:1941-1953. http://doi.org/10.1111/ddi.12991

Dormann, C. F., J. Elith, S. Bacher, C. Buchmann, G. Carl, G. Carré, J. R. García-Marquéz, B. Gruber, B. Lafourcade, P. J. Leitão, T. Münkemüller, C. McClean, P. E. Osborne, B. Reineking, B. Schröder, A. K. Skidmore, D. Zurell, and S. Lautenbach. 2012. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 35:001-020. https://doi.org/10.1111/j.1600-0587.2012.07348.x

Drake, A., D. R. de Zwaan, T. A. Altamirano, S. Wilson, K. Hick, C. Bravo, J. T. Ibarra, and K. Martin. 2021. Combining point counts and autonomous recording units improves avian survey efficacy across elevational gradients on two continents. Ecology and Evolution 11:8654-8682. http://doi.org/10.1002/ece3.7678

Drapeau, P., A. Leduc, J.-F. Giroux, J.-P. L. Savard, Y. Bergeron, and W. L. Vickery. 2000. Landscape-scale disturbances and changes in bird communities of boreal mixed-wood forests. Ecological Monographs 70:423-444. https://doi.org/10.2307/2657210

Elith, J., C. H. Graham, R. P. Anderson, M. Dudík, S. Ferrier, A. Guisan, R. J. Hijmans, F. Huettmann, J. R. Leathwick, A. Lehmann, J. Li, L. G. Lohmann, B. A. Loiselle, G. Manion, C. Moritz, M. Nakamura, Y. Nakazawa, J. M. Overton, A. T. Peterson, S. J. Phillips, K. Richardson, R. Scachetti-Pereira, R. E. Schapire, J. Soberón, S. Williams, M. S. Wisz, and N. E. Zimmermann. 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29:129-151. https://doi.org/10.1111/j.2006.0906-7590.04596.x

Environment Canada. 2016. Recovery strategy for Canada Warbler (Cardellina canadensis) in Canada. Species at Risk Act Recovery Strategy Series. Ottawa, Ontario, Canada. https://www.sararegistry.gc.ca/virtual_sara/files/plans/rs_canada%20warbler_e_final.pdf

Fink, D., T. Auer, A. Johnston, M. Strimas-Mackey, O. Robinson, S. Ligocki, W. Hochachka, L. Jaromczyk, C. Wood, I. Davies, M. Iliff, and L. Seitz. 2021. eBird status and trends, data version: 2020. Cornell Lab of Ornithology, Ithaca, New York. https://doi.org/10.2173/ebirdst.2020

Flockhart, D., G. W. Mitchel, R. G. Krikun, and E. M. Bayne. 2016. Factors driving territory size and breeding success in a threatened migratory songbird, the Canada Warbler. Avian Conservation and Ecology 11(2):4. https://doi.org/10.5751/ACE-00876-110204

Franklin, J. 2013. Species distribution models in conservation biogeography: developments and challenges. Diversity and Distributions 19:1217-1223. http://doi.org/10.1111/ddi.12125

Goodnow, M. L., and L. R. Reitsma. 2011. Nest-site selection in the Canada Warbler (Wilsonia canadensis) in central New Hampshire. Canadian Journal of Zoology 89:1172-1177. https://doi.org/10.1139/z11-094

Government of Alberta. 2012. Alberta wild species general status listing 2010. Fish and Wildlife Division, Sustainable Development Resources, Alberta, Canada. https://open.alberta.ca/dataset/6d247118-2097-43d5-8585-0d2592a48430/resource/adf89f4d-c6bd-4464-897d-3b8c750e0284/download/srd-sar-alberta-wild-species-general-status-listing-2010.pdf

Government of Canada. 2023. Species at Risk Act. Government of Canada, Department of Justice, Ottawa, Ontario, Canada. http://laws-lois.justice.gc.ca/eng/acts/S-15.3/

Grinde, A. R., and G. J. Niemi. 2016. Influence of landscape, habitat, and species co-occurrence on occupancy dynamics of Canada Warbler. Condor 118:513-531. https://doi.org/10.1650/CONDOR-15-168.1

Guisan, A., and N. E. Zimmermann. 2000. Predictive habitat distribution models in ecology. Ecological Modelling 135: 147-186. https://doi.org/10.1016/S0304-3800(00)00354-9

Haché, S., P. Sólymos, T. Fontaine, E. M. Bayne, S. Cumming, F. Schmiegelow, and D. Stralberg. 2014. Analyses to support critical habitat identification for Canada Warbler, Olive-sided Flycatcher, and Common Nighthawk. Project K4B20-13-0367, Final Report 1 and 2. Boreal Avian Modelling Project, Edmonton, Alberta, Canada. https://web.archive.org/web/20180721191540id_/http://www.borealbirds.ca/files/Technical_Reports/Hacheetal2014.pdf

Hansen, M. C., P. V. Potapov, R. Moore, M. Hancher, S. A. Turubanova, A. Tyukavina, D. Thau, S. V. Stehman, S. J. Goetz, T. R. Loveland, A. Kommareddy, A. Egorov, L. Chini, C. O. Justice, and J. R. G. Townshend. 2013. High-Resolution global maps of 21st-century forest cover change. Science 342: 850-53. https://doi.org/10.1126/science.1244693

Hernández, P. A., C. H. Graham, L. L. Master, and D. L. Albert. 2006. The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography 29:773-785. http:/doi.org/10.1111/j.0906-7590.2006.04700.x

Hirzel, A., G. L. Lay, V. Helfer, C. Randin, and A. Guisan. 2006. Evaluating the ability of habitat suitability models to predict species presences. Ecological Modelling 199:142-152. https://doi.org/10.1016/j.ecolmodel.2006.05.017

Hobson, K. A., and E. Bayne. 2000. Breeding bird communities in boreal forest of western Canada: consequences of “unmixing” the mixedwoods. Condor 102:759-769. https://doi.org/10.1093/condor/102.4.759

Hunt A. R., E. M. Bayne, and S. Haché. 2017. Forestry and conspecifics influence Canada Warbler (Cardellina canadensis) habitat use and reproductive activity in boreal Alberta, Canada. Condor 119:832-847. https://doi.org/10.1650/CONDOR-17-35.1

Imbeau, L., M. Mönkkönen, and A. Desrochers. 2001. Long-term effects of forestry on birds of the eastern Canadian boreal forest: a comparison with Fennoscandia. Conservation Biology 15:1151-1162. https://doi.org/10.1046/j.1523-1739.2001.0150041151.x

Jiménez-Valverde, A., J. M. Lobo, and J. Hortal. 2008. Not as good as they seem: the importance of concepts in species distribution modelling. Diversity and Distributions 14:885-890. https://doi.org/10.1111/j.1472-4642.2008.00496.x

Kirk, D. A., W. Diamond, A. R. Smith, E. Holland, and P. Chytyk. 1997. Population changes in boreal forest birds in Saskatchewan and Manitoba. Wilson Bulletin 109(1):1-194. https://sora.unm.edu/sites/default/files/journals/wilson/v109n01/p0001-p0027.pdf

Lahoz-Monfort, J. J., G. Guillera-Arroita, E. J. Milner-Gulland, R. P. Young, and E. Nicholson. 2010. Satellite imagery as a single source of predictor variables for habitat suitability modelling: how Landsat can inform the conservation of critically endangered lemur. Journal of Applied Ecology 47:1094-1102. http://doi.org/10.1111/j.1365-2664.2010.01854.x

Lampila, P., M. Mönkkönen, and A. Desrochers. 2005. Demographic responses by birds to forest fragmentation. Conservation Biology 19:1537-1546. https://doi.org/10.1111/j.1523-1739.2005.00201.x

Leston, L., F. V. Dénes, T. D. S. Docherty, J. A. Tremblay, Y. Boulanger, S. L. Van Wilgenburg, D. Stralberg, P. Sólymos, S. Haché, K. St. Laurent, R. Weeber, B. Drolet, A. R. Westwood, D. D. Hope, J. Ball, S. J. Song, Steven G. Cumming, E. M. Bayne, and F. K. A. Schmiegelow. 2024. A framework to support the identification of critical habitat for wide-ranging species at risk under climate change. Biodiversity and Conservation 33:603-628. http://doi.org/10.1007/s10531-023-02761-1

Lin, H-Y., A. D. Binley, R. Schuster, A. D. Rodewald, R. Buxton, and J. R. Bennet. 2022. Using science data to help identify threatened species occurrences of known ranges. Biological Conservation 268:109523. https://doi.org/10.1016/j.biocon.2022.109523

Matsuoka, S., P. Sólymos, T. Fontaine, and E. Bayne. 2011. Roadside surveys of boreal forest birds: how representative are they and how can we improve current sampling? Report to Environment Canada, Boreal Avian Modelling Project, University of Alberta, Edmonton, Alberta, Canada. https://web.archive.org/web/20180722020527id_/http://www.borealbirds.ca/files/BAM_Report_on_Roadside_Survey_Bias_for_EC.pdf

McShea, W. J. 2014. What are the roles of species distribution models in conservation planning? Environmental Conservation 41:93-96. http:/doi.org/10.1017/S0376892913000581

Miller, N.A. 1999. Landscape and habitat predictors of Canada Warbler (Wilsonia canadensis) and Northern Waterthrush (Seiurus noveboracensis) occurrence in Rhode Island. Thesis. Department of Natural Resources, University of Rhode Island, Kingston, Rhode Island, USA.

Ontario Ministry of Natural Resources and Forestry. 2020. Report on forest management 2016-2017. Ministry of Natural Resources and Forestry, Toronto, Ontario, Canada. https://files.ontario.ca/mnrf-2016-2017-report-forest-management-en-2020-10-01.pdf

Ortega-Huerta, M. A., and J. H. Vega-Rivera. 2017. Validating distribution models for twelve endemic bird species of tropical dry forest in western Mexico. Ecology and Evolution 7:7672-7686. https://doi.org/10.1002/ece3.3160

Pearson, R. G. 2010. Species’ distribution modeling for conservation educators and practitioners. Lessons in Conservation 3:54-89. https://www.amnh.org/content/download/141368/2285424/file/species-distribution-modeling-for-conservation-educators-and-practitioners.pdf

Peterson, A. T., J. Soberón, R. G. Pearson, R.P. Anderson, E. Martinez-Meyer, M. Nakamura, and M. B. Araújo. 2011. Ecological niches and geographic distributions. Princeton University Press, Princeton, New Jersey, USA. https://doi.org/10.1515/9781400840670

Phillips S. J., R. P. Anderson, and R. E. Schapire. 2006. Maximum entropy modelling of species geographic distributions. Ecological Modelling 190:231-259. http://doi.org/10.1016/j.ecolmodel.2005.03.026

Phillips S. J., M. Dudík, and R. Schapire. 2021. Maxent software for modeling species niches and distributions (version 3.4.1). http://biodiversityinformatics.amnh.org/open_source/maxent/

Potapov, P., X. Li, A. Hernández-Serna, A. Tyukavina, M. C. Hansen, A. Kommareddy, A. Pickens, S. Turubanova, H. Tang, C. E. Silva, J. Amstron, R. Dubayah, J. B. Blair, and M. Hofton. 2021. Mapping global forest canopy height through integration of GEDI and Landsat data. Remote Sensing of Environment 253:112165. http://doi.org/10.1016/j.rse.2020.112165

QGIS.org. 2021. QGIS Geographic Information System (version 3.18). QGIS Association. http://www.qgis.org

Reitsma, L., M. Goodnow, M. T. Hallworth, and C. J. Conway. 2010. Canada Warbler (Cardellina canadensis). In A. Poole and F. Gill, editors. Birds of North America. Cornell Lab of Ornithology, Ithaca, New York, USA.

Roberts, D. R., V. Bahn, S. Ciuti, M. S. Boyce, J. Elith, G. Guillera-Arroita, S. Hauenstein, J. J. Lahoz-Monfort, B. Schröder, W. Thuiller, D. I. Warton, B. A. Wintle, F. Hartig, and C. F. Dormann. 2017. Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure. Ecography 40:913-929. http://doi.org/10.1111/ecog.02881

Rosenberg, K. V., A. M. Dokter, P. J. Blancher, J. Sauer, A. C. Smith, P. A. Smith, J. C. Stanton, A. Panjabi, L. Helft, M. Parr, P. P. Marra, et al. 2019. Decline of the North America avifauna. Science 336:120-124. https://doi.org/10.1126/science.aaw1313

Sauer, J. R., W. A. Link, and J. E. Hines. 2020. The North American Breeding Bird Survey, analysis results 1966–2019: U.S. Geological Survey, data release. U.S. Geological Survey, Reston, Virginia, USA. https://doi.org/10.5066/P96A7675

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-analysis. Canadian Journal of Forest Research 36:1299-1318. https://doi.org/10.1139/x06-017

Schieck, J., M. Nietfeld, and J. B. Stelfox. 1995. Differences in bird species richness and abundance among three successional stages of aspen-dominated boreal. Canadian Journal of Zoology 73:1417-1431. https://doi.org/10.1139/z95-167

Statistics Canada. 2022. Census profile, 2021 census of population. Statistics Canada Catalogue no. 98-316-X2021001. Ottawa, Ontario, Canada. https://www12.statcan.gc.ca/census-recensement/2021/dp-pd/prof/search-recherche/results-resultats.cfm?Lang=E&SearchText=Thunder+Bay

Stralberg, D., S. M. Matsuoka, A. Hamann, E. M. Bayne, P. Sólymos, F. K. A. Schmiegelow, X. Wang, S. G. Cumming, and S. J. Song. 2015. Projecting boreal bird responses to climate change: the signal exceeds the noise. Ecological Applications 25:52-69. http://doi.org/10.1890/13-2289.1

Sullivan, B. L., J. L. Aycrigg, J. H. Barry, R. E. Bonney, N. Bruns, C. B. Cooper, T. Damoulas, A. A. Dhondt, T. Dietterich, A. Farnsworth, D. Fink, J. W. Fitzpatrick, T. Fredericks, J. Jeff Gerbracht, C. Gomes, W. M. Hochachka, M. J. Iliff, C. Lagoze, F. A. La Sorte, M. Merrifield, W. Morris, T. B. Phillips, M. Reynolds, A. D. Rodewald, K. V. Rosenberg, N. M. Trautmann, A. Wiggins, D. W. Winkler, W.-K. Wong, C. L. Wood, J. Yu, and S. Kelling. 2014. The eBird enterprise: an integrated approach to development and application of citizen science. Biology Conservation 169:31-40. https://doi.org/10.1016/j.biocon.2013.11.003

U.S. Geological Survey. 2020. Landsat products: Landsat 8 collection 2 level 1, 30 m resolution, June and August 2018. Courtesy of U.S. Geological Survey. https://doi.org/10.5066/P975CC9B

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management. 47:893-901. https://doi.org/10.2307/3808148

Wells, J., N. Dawson, N. Culver, F. Reid, and S. M. Siegers. 2020. The state of conservation in North America’s boreal forest: issues and opportunities. Frontiers in Forests and Global Change 3:1-18. https://doi.org/10.3389/ffgc.2020.00090

Wells, J., D. Stralberg, and D. Childs. 2018. Boreal forest refuge: conserving North America’s bird nursery in the face of climate change. Boreal Songbird Initiative, Seattle, Washington, USA. https://www.borealbirds.org/sites/default/files/publications/Report-Birds-Climate-Change-Web-Res_1.pdf

Wester, M. C., B. L. Henson, W. J. Crins, P. W. C. Uhlig, and P. A. Gray. 2018. The ecosystems of Ontario, part 2: ecodistricts. Science and Research Technical Report TR-26. Ontario Ministry of Natural Resources and Forestry, Science and Research Branch, Peterborough, Ontario, Canada. https://www.cabidigitallibrary.org/doi/full/10.5555/20193434114

Westwood, A. R., C. Staicer, P. Sölymos, S. Haché, T. Fontaine, E. Bayne, and D. Mazerolle. 2019a. Estimating the conservation value of protected areas in Maritime Canada for two species at risk: the Olive-sided Flycatcher (Contopus cooperi) and Canada Warbler (Cardellina canadensis). Avian Conservation and Ecology 14(1):16. https://doi.org/10.5751/ACE-01359-140116

Westwood, R., A. R. Westwood, M. Hooshmandi, K. Pearson, K. LaFrance, and C. Murray. 2019b. A field-validated species distribution model to support management of the critically endangered Poweshiek skipperling (Oarisma poweshiek) butterfly in Canada. Conservation Science and Practice 2(3):e163. http://doi.org/10.1111/csp2.163

Wijedasa, L. S., S. Sloan, D. G. Michelakis, and G. R. Clements. 2012. Overcoming limitations with Landsat imagery for mapping peat swamp forests in Sundaland. Remote Sensing 4:2595-2618. http://doi.org/10.3390/rs4092595

Wintle, B. A., J. Elith, and J. M. Potts. 2005. Fauna habitat modelling and mapping: a review and case study in the Lower Hunter Central Coast region of NSW. Austral Ecology 30:719-738. http://doi.org/10.1111/j.1442-9993.2005.01514.x

Yates, K. L., P. J. Bouchet, M. J. Calely, K. Mengensen, C. F. Radin, S. Parnell, and A. H. Fielding, A. J. Bamford, S. Ban, A. M. Barbosa, et al. 2018. Outstanding challenges in the transferability of ecological models. Trends in Ecology and Evolution 33:790-802. https://doi.org/10.1016/j.tree.2018.08.001

Ziolkowski Jr., D. J., M. Lutmerding, V. I. Aponte, and M-A. R. Hudson. 2022. North American Breeding Bird Survey dataset 1966–2021: U.S. Geological Survey data release, U.S. Geological Survey, Reston, Virginia, USA. https://doi.org/10.5066/P97WAZE5

Zlonis, E. J., and G. J. Niemi. 2014. Avian communities of managed and wilderness hemiboreal forests. Forest Ecology and Management 328:26-34. http://doi.org/10.1016/j.foreco.2014.05.017