Introduction

Wetlands are a dominant feature of the boreal landscape with over 20% of the world’s wetlands occurring in Canada’s boreal zone (Webster et al. 2015). Boreal wetlands are globally important for climate regulatory services due to their ability to reduce air temperatures, sequester carbon and mediate greenhouse gas emissions; they also provide important water regulating services by attenuating downstream flooding, and filtering and retaining pollutants (Millennium Ecosystem Assessment 2005). These aquatic ecosystems are ecologically linked to their surrounding terrestrial watersheds (Hynes 1975) wherein the structure and function of wetlands are threatened by both disturbances to the waterbodies themselves and by disturbances to their upstream watersheds (Kreutzweiser et al. 2013). In the Canadian boreal zone, freshwater ecosystems are under high cumulative stress from climate change, expanding human populations, and resource development (Schindler and Lee 2010).

The Oil Sands Region (OSR) in northern Alberta, Canada covers approximately 140,000 km2 of Alberta’s land area and contains the third largest oil reserves (10.3%) in the world with approximately 166 billion barrels (Government of Canada 2019). Twenty percent of this resource is extracted through surface mining over 3% of the OSR (4,750 km2) in the Athabasca Oil Sands Region (AOSR; ~93, 260 km2). The remainder, and vast majority, of the reserve is located too deep for surface mining and extracted with in situ technologies (Government of Alberta 2024a, b). Bitumen production (both mined and in situ) totalled approximately 3.3 million barrels per day in 2022 (Government of Alberta 2024a, b). Oil sands production in northeastern Alberta has significant potential to affect aquatic ecosystems both directly and indirectly via land disturbance (Ficken et al. 2019), hydrologic alteration (Volik et al. 2020) and contaminants (Mundy et al. 2019a, b; Mundy et al. 2019a, b).

Globally, economic development is one of the primary indirect drivers of wetland degradation and loss (Millennium Ecosystem Assessment 2005). In regions experiencing pressure from development, monitoring and management are critical to maintain the ecological integrity of freshwater resources. Benthic invertebrates are commonly used in bioassessment of freshwaters in North America, including rivers (Kenney et al. 2009), lakes (Parsons et al. 2010a) and wetlands (Tall et al. 2008). Benthic invertebrates spend most of their lives in water and are thus continually exposed to disturbances integrating both episodic and cumulative impacts to water quality and habitat across time and space with species differing in their responses to these disturbances (Barbour et al. 1999). Successful bioassessment programs using benthic invertebrates as monitoring endpoints have been developed for rivers across the globe; however, the development of similar assessment programs in wetlands has shown less progress. Some examples of regional bioassessment programs developed for wetlands include coastal wetlands in the Great Lakes (Burton et al. 1999; Uzarski et al. 2004), depressional wetlands in Minnesota (Gernes and Helgen 2002), and urban wetlands in northern California (Lunde and Resh 2012). In the surface mineable regions of Alberta, previous research has shown that contaminant concentrations are highest near facilities and decrease with distance (Landis et al. 2012; Wieder et al. 2016). Thus, invertebrate assemblages are expected to be more heavily impacted closer to oil sands facilities than those from areas further from facilities. Invertebrate-based bioassessment in the OSR will add an additional line of evidence to support comprehensive long-term monitoring of wetlands in this globally important ecosystem.

The overall objectives of the Oil Sands Monitoring Program are to identify i) if change is occurring, and ii) if those changes are due to oil sands development. The Oil Sands Wetland Ecosystem Monitoring Program (Mahoney et al. 2023), hereafter Wetlands Program began its pilot phase in 2017. Here we present the results of the pilot phase work on water quality and benthic communities of open water wetland habitats. The main objectives of this study are i) to compare water quality parameters of concern between Nearfield (< 5 km from OS facilities) and Farfield (> 5 km from facilities) sites, ii) to compare wetland invertebrate communities between Nearfield and Farfield sites, and iii) to identify what water quality parameters are associated with benthic invertebrate community variability among all sites. Oil Sands Monitoring is an adaptive monitoring program, and the present work is intended to advance wetlands biomonitoring design of the Wetlands program. This paper analyses the first four years of data to recommend improvements to implementation of a full-scale Wetlands Program.

Methods

Study area

Nineteen open water wetlands (i.e., open water zone covering more than 25% of the total area, and a maximum depth of 2 m; (Alberta Environment and Sustainable Resource Development 2015)) across the AOSR (Fig. 1) were chosen to represent a gradient of local land disturbance and distance from oil sands development operations. Thirteen of the wetlands are natural, and the remaining six have been modified by construction (2) or are borrow pits (4). All borrow pits possessed a naturalized littoral margin, and none contained oil sands process-affected materials (e.g., petroleum coke, tailings). Wetlands were separated into two categories (hereafter called “groups”)—Nearfield (NF) and Farfield (FF)—based on previous work that identified significant land disturbances and loss in vegetation canopy coverage within 5 km of oil sands mine boundaries ((Chasmer et al. 2021) Alberta Environment and Parks, Unpublished Data). Seven of the sites with a total of 18 visits were assigned to the Nearfield (NF) group (i.e., < 5 km from oil sands surface mining operations) and 12 sites with a total of 19 visits were assigned to the Farfield (FF) group (i.e., > 5 km from oil sands surface mining operations) (Table 1). Farfield sites are not within close proximity to oil sands surface mining operations but also do not represent true reference condition because they may still be within the deposition radius (~30 km) for fugitive dust (Landis et al. 2012) or near linear features. From 2017 to 2021 (excluding 2020), sites were sampled for water quality and benthic invertebrate community composition. In 2017 and 2018, benthic invertebrate and water quality samples were collected in mid-August; in 2019 and 2021 samples were collected mid-July. Three of the sites had repeat visits in all sampling years, two sites were visited three times, five wetlands were visited twice, and the remaining nine were sampled once (Table 1).

Fig. 1
figure 1

Map of study sites in the Athabasca Oil Sands Region, northeastern Alberta, Canada. Samples were collected annually from 19 wetlands over four years (2017–2021, excluding 2020). Surface mining map layer provided by Alexander and Chambers (2016)

Full size image
Table 1 Location, characteristics (years sampled, group, HFI (%), and history) of 19 wetland monitoring sites in the Athabasca Oil Sands Region, northeastern Alberta, Canada
Full size table

Environmental co-variables

Water chemistry samples were collected following standard collection protocols for monitoring of surface waters in Alberta (Government of Alberta 2006). A single representative grab sample was collected from shoreline at mid-water depth at each site. Samples for water quality were collected prior to any other sampling to minimize disturbance of the site and to limit particulate organic matter in the water column. Samples were collected to analyse a modified suite of water quality parameters as outlined in the Lower Athabasca Water Quality Monitoring Program Phase 1 (Wrona and di Cenzo 2011). Briefly, this includes routine parameters (e.g., specific conductance, pH), major ions, nutrients and carbon, trace elements including mercury, and polycyclic aromatic compounds. Samples were preserved on site, where required, and kept in a cold storage prior to analysis at accredited commercial laboratories.

Human footprint (HF) was calculated as percent of total human footprint in a 500 m buffer around the wetland edge using the Alberta Human Footprint Monitoring Program Inventory (Government of Alberta 2019). Human footprint was used as a measure of land disturbance in the wetland periphery and includes several disturbance categories related to energy and mining, forestry, and transportation. In the AOSR, between 2008 and 2018, disturbed wetland area was primarily attributed to oil and gas (63.1%), with an additional 26.6% and 5.7% attributed to forestry and to transportation, respectively (Montgomery, Mahoney et al. submitted). Over the course of monitoring there were no major changes in land disturbance at repeat sampling sites apart from ATH02, where a portion of forest west of the wetland was logged between 2018 and 2019. This increased the HF from 1.4% to 12.1%. Distance to nearest oil sands mine was calculated as the distance from the sampling location to the nearest active open-pit mine fence line (in situ operations were out of scope for the pilot phase of the program). Distance to nearest mine is a proxy for general mining impacts, especially those associated with aerial deposition of fugitive dust.

Benthic invertebrates

Samples of the benthic invertebrate community were collected following the national standard Canadian Aquatic Biomonitoring Network (CABIN) protocol for wetland habitats (Armellin et al. 2019). A single, continuous sweep was conducted in the emergent and submerged vegetation zones using a 400 µm kick net for a total of two minutes. Samples were collected from the littoral zone in an area considered most representative of the wetland after a visual assessment was conducted. Ideally, samples should be collected while wading through the sample area, but some samples were collected while moving along the edge of the wetland due to unstable sediment. Benthic samples were collected close to the location of the water sample, but in an area not previously disturbed by sample collection. Excess vegetation was rinsed on site for no more than 20 min, and then discarded. Samples were preserved in 95% ethanol and kept in cold, dark storage.

Invertebrate samples were sent to a certified taxonomic laboratory for identification and enumeration. Benthic invertebrate samples were processed according to the CABIN standard (McDermott et al. 2014) including QA/QC procedures. Each benthic sample was subsampled using either a Marchant box or Caton tray. A minimum of 5% of each sample was processed. If after 5% the sample did not include 300 individuals, then subsampling continued until 300 individuals were identified. Certified taxonomists identified invertebrates to the lowest practical taxonomic level (usually genus). This procedure generates abundance estimates suitable for assessing compositional change associated with exposure to environmental stressors.

Statistical analysis

Fifteen water quality parameters were chosen a priori to compare between Nearfield and Farfield groups. These included standard water quality variables (pH, specific conductance, turbidity), total phosphorus, and variables previously shown to be associated with oil sands development activities. Total nitrogen (N), and dissolved sulphate (SO42−) were included as potential stressors associated with stack emissions of NOx and SOx. Base cations were chosen as they are broadly associated with fugitive dust deposition from surficial erosion (Landis et al. 2012), while vanadium and nickel (petrogenic), and aluminum and iron (crustal) were chosen as tracers for mining activity (Landis et al. 2019). Polycyclic aromatic compounds (including parent hydrocarbons, alkylated hydrocarbons and dibenzothiophenes) were chosen as stressors associated with bitumen deposits and petroleum coke (Zhang et al. 2016). Methylmercury was included due to concerns of wetlands acting as methylation hot spots (Wasiuta et al. 2019), and biomagnification in the aquatic food web (Lavoie et al. 2013). Where parameters were below the detection limit of the analytical method, one-half of the reported detection limit was used in analyses.

Water quality parameters were standardized to a mean of zero and a standard deviation of one. The assumption of multivariate normality could not be satisfied so permutational multivariate analysis of variance (PERMANOVA (Anderson 2017); function: adonis2; package: vegan; Oksanen et al. 2017) with restricted permutations blocked by Year (function: how; package: permute) was used with Euclidean distance to compare all water quality variables between groups simultaneously. After confirming a significant difference between groups, pairwise comparisons were done to determine which parameters differed specifically. Parameters were log-transformed to meet assumptions of normality and compared with Analysis of Variance (ANOVA (Fisher 1921); function: aov; package: stats). For three parameters (P, SO42−, Al), transformed data was not normally distributed so a non-parametric Wilcoxon rank test (function: wilcox.test; package: stats) was used. Benjamini-Hochberg (Benjamini and Hochberg 1995; Waite and Campbell 2006) was used with a false discovery rate of 0.20 (Lee and Lee 2018) to confirm that significant results were not a result of Type I error from multiple comparisons.

Benthic invertebrate data were summarised per site in a matrix of total abundance at family level (subfamily for chironomids). Assemblage level patterns are typically similar whether using coarse (family) or fine (genus) taxonomic resolution (Pires et al. 2021), and so family level was used here to include more individuals (e.g., juveniles that cannot be identified to genus) from the dataset. There were 63 Families across all sites and monitoring years. Rare families present in fewer than 5% of samples (i.e., present in only one sample) were removed from the data set prior to analysis (Boersma et al. 2016; Gleason and Rooney 2017). Thirteen families meeting this criterion were removed: Aturidae, Hygrobatidae, Mideidae and Teutoniidae (Order: Trombidiformes), Belostomatidae and Gerridae (Order: Hemiptera), Cordulegastridae (Suborder: Anisoptera), Ephydridae (Order: Diptera), Gerridae and Hydrophilidae (Order: Coeloptera), Hydrobiidae (Order: Littorinimorpha), Lepidostomatidae (Order: Trichoptera) and Siphlonuridae (Order: Ephemeroptera). Additionally, juvenile or damaged individuals identified only to order level (or higher) were excluded from multivariate analysis (i.e., 1,690 rare and/or damaged individuals were removed comprising 4% of the total number of individuals collected).

Beta diversity between Nearfield and Farfield groups was analysed using the test for multivariate homogeneity of group dispersions (Anderson 2006; function: betadisper; package: vegan). In this regard, beta diversity was measured as the average distance (i.e., dissimilarity) from an individual unit to the group spatial median. Dispersion is high when communities are heterogeneous among sites within a group (i.e., NF vs FF). In addition to measuring dispersion, differences in location were analysed using PERMANOVA with restricted permutations blocked by Year. This tests for location differences in multivariate space between group centroids where distance is defined by Bray–Curtis dissimilarity.

Detrended Correspondence Analysis (DCA (Hill and Gauch 1980)) was performed to determine if the gradient response was linear or unimodal (function: decorana; package: vegan). The first axis gradient length was < 2.5 (Legendre and Legendre 1998) so RDA was chosen for a linear constrained analysis. Redundancy analysis (RDA (van den Wollenberg 1977); function: rda; package: vegan) performs multiple regression comparing continuous explanatory variables to a community ordination. It evaluates to what extent variation in community composition among samples is redundant with variation in the chosen environmental covariates. Forward selection (function: ordiR2step; package: vegan) following Blanchet et al. (2008) was used to avoid overfitting the model and determine which parameters (if any) had a significant contribution to the variance explained by the global model.

For the global RDA model, the previous 15 water quality variables were used as well as human footprint (HF) in the surrounding 500 m buffer (a measure of broad land disturbance), and distance to nearest oil sands mine (as a proxy measure for deposition load). Prior to the analysis, multicollinearity in the environmental dataset was evaluated by calculating variance inflation factors (function: vif.cca; package: vegan). Specific conductance (53.3) and base cations (54.8) had high (> 10 Dormann et al. 2013) variance inflation and were significantly correlated (r = 0.98, p < 0.001) so base cations were removed from the dataset.

All analyses were performed using R version 4.1.2 statistical software (R Core Team 2021).

Results

Water quality

There was a significant difference in overall water quality between NF and FF groups (F = 2.38, p = 0.036). The concentrations of specific conductance (Fig. 2A; F = 7.99, p = 0.008), dissolved sulphate (Fig. 2F; W = 241.5, p = 0.024), dissolved iron (Fig. 3B; F = 6.52, p = 0.016), and total dibenzothiophenes (Fig. 4C; F = 14.96, p = 0.0004) were significantly higher in NF sites after B-H correction. Median vanadium concentrations (Fig. 3C; F = 3.25, p = 0.08) were higher in NF sites, but the difference was non-significant. Nearfield sites had more outliers near or exceeding long-term Protection of Aquatic Life guidelines (50 µg L−1 at pH ≥ 6.5 (Government of Alberta 2018)) for dissolved aluminum (Fig. 3A), but there was no difference in group medians (W = 131, p = 0.23). For all remaining parameters, there were no significant differences after B-H correction.

Fig. 2
figure 2

Boxplots (minimum, median, maximum, interquartile range, and outliers) showing differences in concentration of standard water quality variables and nutrients between Nearfield and Farfield wetlands. Specific conductance (Panel A; F = 7.99, p = 0.008) and dissolved sulphate (Panel F; W = 241.5, p = 0.024) were significantly higher in the Nearfield group

Full size image
Fig. 3
figure 3

Boxplots (minimum, median, maximum, interquartile range, and outliers) showing differences in trace inorganics concentrations from Nearfield and Farfield wetlands. Dissolved iron (Panel B; F = 6.52, p = 0.016) was significantly higher in the Nearfield group. Dashed red line shows the long-term Protection of Aquatic Life guidelines

Full size image
Fig. 4
figure 4

Boxplots (minimum, median, maximum, interquartile range, and outliers) showing differences in concentration of A) sum of parent PAHs, B) sum of alkylated PAHs, and C) sum of DBTs between Farfield and Nearfield wetland groups. Concentration of dibenzothiophenes was significantly higher (F = 14.96, p = 0.0004) in Nearfield wetlands

Full size image

Invertebrate community

The distance to spatial median was significantly higher (F = 11.2, p = 0.003) in FF wetlands compared to NF wetlands (Fig. 5A), but there was no difference in multivariate location (F = 1.89, p = 0.118) between groups (Fig. 5B).

Fig. 5
figure 5

Boxplot (minimum, median, maximum, interquartile range, and outliers) and NMDS ordination of benthic invertebrate community composition showing differences between groups (FF and NF) in A) dispersion and B) location. Farfield wetlands are represented by purple circles, and Nearfield sites represented by gold triangles. Beta diversity was significantly higher (F = 11.2, p = 0.003) in Farfield wetlands. NMDS stress is 0.15

Full size image

The total adjusted variance explained by the global RDA model was 23.9%. Only specific conductance (11.2%, p = 0.01) and pH (4.9%, p = 0.03) were significant drivers of invertebrate community variability. The next candidate parameter was total vanadium, but it failed significance testing (p = 0.09) for model inclusion. Compositional differences among communities were unrelated to group (NF versus FF). Most sites were clustered around the origin, and there is no separation along axes 1 or 2 between Nearfield and Farfield sites (Fig. 6A). Family Caenidae was correlated with high specific conductance, while Family Naididae was correlated with low pH (Fig. 6B).

Fig. 6
figure 6

Redundancy analysis biplots projecting RDA axes 1 and 2 showing A) wetland sites, and B) benthic invertebrate Families (or Subfamily; chironomids) from the Athabasca Oil Sands Region, and water quality parameters (specific conductance and pH) driving variation in benthic invertebrate community composition. Farfield wetlands are represented by purple circles and Nearfield sites represented by gold triangles. The total adjusted variance explained by the RDA was 23.9% with 11.2% explained by SPC and an additional 4.9% explained by pH

Full size image

Discussion

Water quality

Concentrations of dissolved sulphate were significantly higher in Nearfield sites, whereas most Farfield sites were below the method detection limit (typically 1.0 mg L−1). The main source of sulphur deposition is emissions of sulphur dioxide from upgrader stacks (Proemse et al. 2012; Fenn et al. 2015). Other studies have confirmed sulphur deposition is highest near the industrial centre (Hsu et al. 2016), and sulphate concentrations in wetlands decrease with increased distance from the industrial centre (Wieder et al. 2016). In addition to atmospheric deposition, groundwater intrusion is an alternate pathway for sulphate transport to wetlands. Higher sulphate concentrations are common in tills in northeastern Alberta (Birks et al. 2019, Birks et al. 2022). Along with the naturally elevated sulphate in regional groundwater, seepage from tailing ponds is a second potential pathway for sulphate to enter groundwater. Processing of bitumen ore produces bitumen froth with elevated levels of pyrite (FeS2) that when exposed to moisture and air oxidizes producing high concentrations of H+, sulphate and soluble iron (Lindsay et al. 2019; Xu 2021). More work is needed to identify areas of groundwater intrusion to open water wetlands in the region and to determine to what extent (if any) tailings seepage is contributing to elevated sulphate and iron to Nearfield monitoring sites. Sulphate is a concern for aquatic life due to direct toxicity (Elphick et al. 2011) and its acidifying effects (Courtney and Clements 1998), but its acidifying effects may be ameliorated by concurrent deposition of base cations (Watmough et al. 2014; Makar et al. 2018).

The median total vanadium concentration was higher in NF sites though not significantly different from FF sites. Previous studies have shown vanadium enrichment in NF floodplain lake sediments downstream of oil sands mine operations (Klemt et al. 2020). Thus, further sampling of NF and FF sites is recommended to assess if vanadium enrichment may be occurring in NF wetlands.

Dibenzothiophenes (DBTs) were higher in NF wetlands, and these are considered indicative of industrial activities in the region. Several environmental media (air, snow, sediment) have shown declines in DBTs concentrations with increasing distances from oil sands producers with no non-petroleum source identified (Harner et al. 2018). There was no significant difference in parent or alkylated polycyclic aromatic hydrocarbons, and the median concentration of alkylated compounds was higher in Farfield sites. Alkylated hydrocarbons are generally associated with byproducts of the upgrading process (i.e., petroleum coke Zhang et al. 2016; Xu 2018)) but are also released from vegetation during wildfires (e.g., retene (Zhang et al. 2022).

Benthic invertebrates

Higher heterogeneity of the benthic invertebrate community in FF sites suggests higher biodiversity in wetlands outside of the industrial centre. However, it may also be due to a wider gradient of habitat conditions. The sites near surfacing mining are all mineral wetlands. Sites far from surface mining include both mineral wetlands and shallow boreal lakes with organic substrate. The NF group represent a subsample of the full spectrum of wetlands included in the FF group, which may explain the wide overlap in location between groups as well as the lower dispersion in the Nearfield group.

The areas of the landscape where surficial glaciated deposits are shallow and bitumen deposits are shallow enough to access via surface mining also support shallow open water mineral wetlands. Sites further afield with deep glaciated substrates overlying deep bitumen deposits precluding surface mining, tend to be large peatland complexes that may include shallow open water wetlands overlying peat deposits. This highlights one of the major challenges in monitoring wetlands in the region: lack of comparable wetland habitat across the landscape in areas near to vs far from the industrial oil sands mining centre.

Factors driving patterns in the benthic communities

Specific conductance was significantly higher in NF sites and explained the most variance in invertebrate communities among sites. Similar results have been documented previously in NF wetlands (Whelly 2000). Most sites characterized by higher specific conductance had land disturbances or were close to open-pit mining operations, and as such, it is likely that the variation in conductivity among sites is associated with human development. Although determining the precise source or sources (e.g., impacts from mines, road development, pipeline construction, etc.) is outside of the scope of this study, all the affected sites are adjacent to roads (e.g., Highway 63, industry access roads). Thus, measured increases in conductivity are likely due to a combination of run-off and/or deposition of fugitive dust from roads. Benthic invertebrates have been shown to respond to changes in water quality driven by land use changes in the watershed (Cooper et al. 2006). There is also a possibility that natural variation in conductivity among sites is associated with groundwater connectivity that may be driving some of the patterns seen in the benthic communities. There are several areas across the AOSR where groundwater flows through salt-rich geologic deposits before discharging to the surface, which creates pockets of naturally saline environments including a saline fen (Wells and Price 2015) south of Fort McMurray and the La Saline Natural Area north of town. However, these areas typically have much higher conductivity (e.g., 10,000 to 20,000 µS cm−1) than was measured in this study. Further work on groundwater-surface water exchange is needed to determine to what extent high conductivity at wetland monitoring sites is due to natural vs anthropogenic factors.

Although a few invertebrate families were correlated with increasing conductivity, Caenidae (Order: Ephemeroptera) showed the strongest response of all taxa. Caenidae at these sites were exclusively in the genus Caenis with C. youngi (Roemhild 1984) as the most common species. Caenis is prevalent in both lentic and lotic waters in a variety of habitats and substrate types in the Interior Plains of Alberta (Corkum 1989). It is ubiquitous and often found in extremely high abundance. In addition to Caenidae, non-insect taxa (e.g., gastropod family Planorbidae) were common at high conductivity sites. Similarly, work on coastal wetlands of the Great Lakes found gastropods to be more abundant in impacted versus reference wetlands (Kashian and Burton 2000), although this is not solely driven by high conductivity.

pH also explained a significant, but low amount of variance among sites. pH is a common water quality parameter affecting biota and was found to be important for lakes in the region (Parsons et al. 2010a, b). Like the Parsons et al. (2010a, b) study, Family Naididae (Subclass: Oligochaeta) showed a strong negative correlation with pH. Decades of research on acid rain has established the effects of acidification on aquatic ecosystems (Cowling and Linthurst 1981), but this is likely a low risk for the region. Despite a prevalence of acid sensitive habitats and poorly buffered soils, modeling shows limited potential for acidification impacts (Whitfield et al. 2010).

Water quality is only one habitat factor with the potential to drive the composition of biological communities. Unexplained variance in the invertebrate community may be due to substrate type (Sanders 2000; Starr et al. 2014), vegetation composition (Gleason et al. 2018; Hanisch et al. 2020, 2023), and/or the presence/absence of fish (Hentges and Stewart 2010; Hanisch et al. 2023). Determining the baseline variability in wetlands across the region will be important for identifying changes to the ecosystem, and determining to what extent these changes are associated with oil sands development.

Our work is among the first published looking explicitly at the responses of invertebrate communities in natural, off-lease wetlands in the Athabasca Oil Sands Region. Despite the prevalence of wetlands on the landscape, most of the work to date has focused on lakes and rivers—both the mainstem Athabasca and its tributaries. Previous assessments on rivers found that benthic macroinvertebrate assemblages in the mainstem Athabasca differ from reference areas (Culp et al. 2018) and are affected by both bitumen-derived contaminants (Gerner et al. 2017) and nutrient enrichment from sewage effluent (Culp et al. 2020). This work has determined that aquatic invertebrates in the OSR can respond to anthropogenic stressors, including those likely from oil sands operations; however, not all areas of the OSR—especially near in situ facilities in the Peace River and Cold Lake Regions—have abundant erosional river habitat suitable for CABIN sampling with the wadeable streams method (Environment Canada 2012). Thus, a wetland bioassessment program to detect potential oil sands impacts could be especially useful in regions like these with ample wetland habitat, but relatively rare erosional lotic habitat suitable for sampling.

However, there are a few potential caveats to consider for bioassessment using invertebrates from lentic systems in the oil sands region. Wetland invertebrates may be relatively less sensitive to the stressors of concern than those of lotic habitats, since they are adapted to harsh conditions with dramatic seasonal (and diel) variations in hydrology, oxygen availability, and temperature (Cooper et al. 2009). For example, lake assessments in the AOSR have shown increased deposition of contaminants (e.g., dibenzothiophenes), but shifts in lake communities are instead associated with warming (Summers et al. 2017). However, a recent study in open water wetlands in the Peace-Athabasca Delta, downstream of oil sands mining development detected significant macroinvertebrate community responses to variation in flood and thermal regimes (Bush et al. 2020) indicating benthic invertebrates can be reliable indicators of differences in water quality conditions.

Wetland water quality differs in sites near surface mines compared to sites further afield of the main development corridor. These differences should be investigated further by comparing true reference condition sites (i.e., sites outside of both direct and indirect impacts from mining activities) in the region to increase the ability of detecting source-effect relationships that meet the objectives of the Oil Sands Monitoring Program. Some evidence suggests that benthic invertebrates respond to variability in water quality, but not necessarily those associated with direct mining impacts. Rather, the relevant parameters may be indirectly associated with mining activities such as runoff and fugitive dust. As wetland monitoring in the region progresses, it should focus on increasing the number of monitoring sites to improve statistical power, account explicitly for differences in wetland habitat type (e.g., mineral vs organic wetlands), and work towards developing region-specific wetland metrics of habitat condition.