Open access

Are different benthic communities in Arctic delta lakes distinguishable along a hydrological connectivity gradient using a rapid bioassessment approach?

Publication: Arctic Science
6 August 2020

Abstract

Aquatic habitats in the Canadian Arctic are expected to come under increasing stress due to projected effects of climate change. There is a need for community-based biomonitoring programs to observe and understand the effects of these stressors on the environment. Here we present results from a 5 year annual sampling program of benthic invertebrates from lakes in the Mackenzie Delta, Northwest Territories, using a rapid bioassessment protocol. Connectivity between the deltaic lakes and main channels is a major driver of lake function and is expected to be substantially impacted by climate change. Lakes were selected along a gradient of connectivity based on sill elevation above the river. Using multivariate analyses of community structure, we determined that benthic assemblages responded to differences in connection time among lakes. This response was detected using a coarse taxonomic level that could be applied by community groups or volunteers but was stronger when invertebrates were identified to the family and genus levels. A secondary gradient was observed that corresponded to productivity gradients in lakes that are isolated from the river during summer. We show that benthic assemblages have potential use as sensitive indicators of climate-mediated changes to the hydrology of lakes in the Mackenzie Delta.

Résumé

Les habitats aquatiques de l’Arctique canadien devraient subir un stress croissant en raison des effets prévus des changements climatiques. Il faut des programmes communautaires de biosurveillance pour observer et comprendre les effets de ces facteurs de stress sur l’environnement. Nous présentons ici les résultats d’un programme d’échantillonnage annuel quinquennal des invertébrés benthiques des lacs du delta du Mackenzie, dans les Territoires du Nord-Ouest, au moyen d’un protocole de bioévaluation rapide. La connectivité entre les lacs deltaïques et les principaux chenaux est un facteur important de la fonction des lacs et devrait être considérablement affectée par les changements climatiques. Les lacs ont été choisis le long d’un gradient de connectivité fondé sur la hauteur du seuil au-dessus de la rivière. À l’aide d’analyses multivariées de la structure des communautés, nous avons déterminé que les assemblages benthiques répondaient aux différences de temps de connexion entre les lacs. Cette réponse a été détectée à l’aide d’un niveau taxonomique général pouvant être appliqué par des groupes communautaires ou des volontaires, mais elle était plus forte lorsque les invertébrés étaient identifiés aux niveaux de la famille et du genre. On a observé un gradient secondaire qui correspondait aux gradients de productivité dans les lacs isolés de la rivière pendant l’été. Nous montrons que les assemblages benthiques peuvent être utilisés comme indicateurs sensibles des changements climatiques dans l’hydrologie des lacs du delta du Mackenzie. [Traduit par la Rédaction]

Introduction

The western Canadian Arctic has experienced increases in mean annual temperature of 2–3 °C and in winter temperature of 3–4 °C over the past six decades, which are projected to rise a further 3–4 °C and 7 °C, respectively, by the end of the 21st century (ACIA 2005). Climate change is likely to have effects on the water balance (MacDonald et al. 2017), hydrology (Krogh and Pomeroy 2018; Lafrenière and Lamoureux 2019; Lininger and Wohl 2019), productivity (Kendrick et al. 2018), and community structure (Laske et al. 2016; Klobucar et al. 2018) of aquatic ecosystems, but considerable uncertainty exists in projections of factors such as precipitation, discharge, evapotranspiration rates, and ice dynamics, such that the direction and magnitude of ecosystem changes are difficult to predict (ACIA 2005). Prediction of the ecological effects of these physical changes is more difficult still due to the complexity of ecological interactions (Kharouba et al. 2018) and region- or lake-specific differences in trajectories of change (Rouse et al. 1997; Lesack and Marsh 2010; Vonk et al. 2015). Given this uncertainty there is a need for ecosystem monitoring tools that use bioindicators that are integrative (i.e., respond to ecological changes along multiple gradients) and capable of responding to environmental change on a suitably rapid time scale.
The composition of biotic communities is widely used to assess the ecological state of a habitat in relation to its environmental characteristics. Bioassessment programs can be used to detect impairments in ecological function and changes to community structure caused by altered environmental gradients, e.g., climate change, altered flow regimes, or anthropogenic pollution. Benthic macroinvertebrate (BMI) communities are widely used in bioassessment programs as indicators of freshwater habitat quality (Rosenberg and Resh 1993). Many programs incorporate rapid bioassessment techniques in which field surveys can be conducted by non-experts (e.g., community or volunteer groups; Thornhill et al. 2019) and can provide benthic community count data with a minimum of processing (Borisko et al. 2007). The Canadian Aquatic Biomonitoring Network (CABIN) protocol for stream bioassessment has been deployed across the Northwest Territories to establish baseline conditions prior to gas pipeline development, as well as by community-based monitoring programs supported by the Cumulative Impacts Monitoring Program of the Government of the Northwest Territories and frequently supported by local Indigenous organizations (Government of Northwest Territories 2018). Due to the logistical difficulty and expense of working at remote northern sites most surveys of BMI communities in the Canadian Arctic conducted by southern researchers have been based on single sampling events, which can bias results towards short-term conditions (Huttunen et al. 2018), and efforts to conduct annual sampling are generally limited to areas near towns that can be accessed relatively easily (e.g., Medeiros et al. 2011; the present study). The use of a rapid bioassessment approach (RBA) that can be carried out by non-experts that have better access to sites would therefore be of great use in generating repeated sampling of remote areas.
In recent years there has been increased understanding of the role of hydrological connectivity in shaping the BMI communities in floodplain systems (Funk et al. 2017). Hydrological connection among sites determines many aspects of the physicochemical habitat template including disturbance, water balance, and nutrient inputs (Junk et al. 1989). In semi-isolated water bodies, connectivity determines the accessibility of a site to organisms with entirely aquatic life cycles, which limits the occurrences of some BMI (e.g., mollusks and crustaceans) directly (Petsch et al. 2017) as well as the occurrence of fish (Laske et al. 2016; Scarabotti et al. 2017), which may influence BMI composition by top-down control (e.g., Ruetz et al. 2002; González-Bergonzoni et al. 2014). Much of this research has focused on lateral hydrological connectivity gradients within the floodplains of major river corridors in temperate or tropical regions, in which flooding may occur multiple times a year and is driven by increases in upstream discharge (Tockner et al. 2000; Gallardo et al. 2014). To our knowledge, no such studies have examined BMI communities in northern systems with ice-jam-driven flood regimes.
The Mackenzie Delta is the largest outlet to the Arctic Ocean in North America and an important habitat for many fish, mammals, and migratory birds. Between the main channels, over 50% of the land area is occupied by ∼49 000 small deltaic lakes (Emmerton et al. 2007), which depend on annual spring flooding from the river to maintain water and sediment balances, refresh the lake water, and provide dispersal routes among lakes for aquatic organisms. Spring flooding is the major annual hydrological event, and sedimentation and biogeochemical transformation of the river water during inundation alters the quantity and quality of water and sediments exported to the Beaufort Sea (Emmerton et al. 2008). Lakes are perched over the main channels at varying elevations, which determines the extent to which they flood during peak water levels, which occur once per year in late May–early June (Marsh and Hey 1989). Overbank flooding in the delta is caused by ice jams that form when mechanical breakup precedes thermal melting of the ice (Beltaos 2013), so that the extent and duration of hydrological connections are likely to be highly sensitive to changes in air temperature during the spring (Lesack et al. 2013, 2014). Previous surveys of BMI communities in the Mackenzie Delta have been rare and were related to environmental assessment of possible pipeline development in the 1970s (e.g., Rosenberg and Snow 1975; Wiens et al. 1975).
The aim of the present paper is to determine the usefulness of BMI communities as rapid bioindicators of lake environments along a gradient of hydrological connectivity. We do this by (1) classifying the sites based on BMI assemblage composition using a non-hierarchical clustering method, (2) identifying environmental gradients that explain community variation, and finally (3) evaluating the fit between BMI-based site classification and the important environmental gradients. Previous work in the Mackenzie Delta has shown the duration of river–lake connection to be a dominant factor determining the chemical composition (Lesack et al. 1998; Tank et al. 2011), primary productivity (Squires et al. 2009), and metabolic balance (Tank et al. 2009a, 2009b) of deltaic lakes in the region, and susceptible to alterations due to climate change (Lesack et al. 2013, 2014) with likely effects on habitat quality and diversity in the Delta (Lesack and Marsh 2010). BMI communities and habitat conditions were sampled in a set of lakes near the town of Inuvik, Northwest Territories, that spanned the range of lake sill elevations present in the study area. We sampled the lakes each summer over five consecutive years that spanned a range of timing and magnitude of peak flood. Here we evaluate the ability of three taxonomic resolutions (a low resolution appropriate to a rapid bioassessment approach, family-, and genus-level) to classify lake communities based on hydrological connectivity and to detect ecological changes within lakes related to changing hydrological conditions.

Methods

Study area

The study area was situated along the East Channel of the Mackenzie Delta near the town of Inuvik (68°21′N, 133°43′W), approximately 90 km downstream of Point Separation where the Mackenzie River enters the delta (Fig. 1). The Mackenzie System drains 1.75 × 106 km of northwestern Canada with an average annual discharge of 316 km3/year, the largest Arctic drainage in North America and globally among the six largest river systems draining into the Arctic Ocean (Holmes et al. 2012). The subarctic climate is characterized by a growing season that is cool and short (14 °C average July temperature) with ice cover from late October until late May, and with low precipitation (∼300 mm/year, Stewart et al. 1998). Due to the region’s high latitude there are extremes in solar radiation with a period of 24-h darkness from early December to early January and 24-h daylight from late June to early July. Continuous permafrost is present throughout the region with active layer depth ranging from 0.3 to 1.5 m (Mackay 1995). The study area is located at the eastern edge of the Delta approximately 10 km south of treeline in the nearby uplands (within the Delta treeline extends approximately 75 km northwest of the study area; Palmer et al. 2012) and is densely covered by coniferous forest, predominantly white spruce (Picea glauca), alder (Alnus crispa) and willows (Salix spp.) (Burn and Kokelj 2009).
Fig. 1.
Fig. 1. Locations of study sites along the East Channel of the Mackenzie Delta near Inuvik, Northwest Territories. Base map data from Natural Resources Canada (https://atlas.gc.ca/toporama/en/index.html). Contains information licensed under the Open Government Licence — Canada.
Within the delta are nearly 45 000 discrete, generally shallow floodplain lakes that cover approximately 25% of the active delta area (Emmerton et al. 2007). The major hydrological event in the Mackenzie Delta is the spring flood, with the increase of the river level and duration of the flood determined by the timing of ice pack breakup in the main channels. Mackay (1963) classified Mackenzie Delta lakes according to their degree of connection to the main channels. Although a portion (approximately 12%) of lakes are connected to the main river channels by perennial distributary channels (No Closure lakes), most lakes are isolated from the river during the ice-free season to a degree determined by the elevation of the sill between the lake outlet and the river channel. Most of these lakes (55%) are inundated during spring flooding, remaining connected to the river for two weeks or longer and referred to as Low Closure lakes. The remaining 33% of lakes have high sill elevations such that they are flooded only for a brief period (up to 2 weeks) in the spring and may have return periods of >1 year (2–4 years at the high end of the range of sill elevations) (Marsh and Hey 1989). These High Closure lakes may have negative water balances due to low precipitation (Marsh and Lesack 1996), and often have physical signs of thermokarst activity (Tank et al. 2011) such as deepening of basins and slumping banks and trees (Bouchard et al. 2017).

Collection methods

Sampling was conducted annually during the summers of 2013–2017 and occurred between 20 July and 10 August of each year to ensure seasonal comparability. The study lakes (Fig. 1) were located close to the town of Inuvik where a hydrometric monitoring station operated by Environment and Climate Change Canada (station 10LC002) provides real-time water level data applicable to the main channels in the study area. Lakes were selected to represent the gradient from High Closure to No Closure lakes according to their sill elevation (Fig. 2a). Sill elevations of the lakes were obtained from the literature (Marsh and Hey 1988) for all but three of the lakes for which historical data were lacking. For these three lakes (MD2, MD3, MD4) sill elevation was interpolated based on the difference between elevations measured lakeside with a Global Positioning System (GPS) unit and the historical sill elevations for the rest of the lakes. In 2013, three connected (No Closure) lakes were sampled (L79A, L80, and L85) but in 2014 and 2015 blockages in the long distributary channel leading to the lakes prevented boat access and these lakes were replaced with two more accessible lakes (L129A and L129B) in 2015–2017.
Fig. 2.
Fig. 2. (a) Spring sill elevation of the study lakes. (b) Peak flood levels at station 10LC002 in the main channel adjacent to the study lakes during the study period. The boxplot shows the distribution of historical flood levels for the station (1985–1990, 2002–2017). m.a.s.l., metres above sea level.
To characterize water quality and aquatic habitat we measured temperature, conductivity, pH, and dissolved oxygen with a YSI (Yellow Springs, Ohio, USA) probe, and collected a 2 L water sample for measurement of dissolved and suspended organics, nutrients, major ions, and chlorophyll (see Table 1). Water samples were sent to the National Laboratory for Environmental Testing (Burlington, Ontario, Canada) for analysis using standardized methods (Environment Canada 2017). The variables were chosen to represent established physical and ecological gradients in the Mackenzie Delta (Lesack et al. 1998; Squires et al. 2009; Tank et al. 2011) and include variables that have been found to structure BMI communities in Arctic regions (e.g., Culp et al. 2019) and (or) deltaic floodplain lakes (e.g., Gallardo et al. 2008).
Table 1.
Table 1. Summary of environmental variables collected from all sites, with the transformation applied for analysis and the scores on the first two principal component (PC) analysis axes.

Note: CT, connection time; POC, particulate organic carbon; PON, particulate organic nitrogen; TDN, total dissolved nitrogen; DIC, dissolved inorganic carbon; TP, total phosphorus; TDP, total dissolved phosphorus; SRP, soluble reactive phosphorus; ChlA, chlorophyll a; DO, dissolved oxygen. Variables selected for redundancy analysis models in at least one benthic macroinvertebrate taxonomic level are in bold.

BMI samples were taken from the littoral zones of the lakes. The average depth for all lake closure types is quite shallow (1.66 m for No Closure, 1.37 m for Low Closure, and 1.84 m for High Closure; Emmerton et al. 2007) and macrophyte growth occurred throughout the basins of most lakes; thus, littoral sampling was chosen because (1) it is representative of the majority of BMI habitat in the study area and (2) the method was applicable to every site regardless of maximum depth. At each sampling event three replicate D-net benthic samples were collected at stations dispersed along the littoral zone. For each D-net sample, approximately 1 m2 of substrate was kicked and suspended while sweeping with a 500 μm mesh D-net for 3 min. Samples were taken approximately 1 m from shore unless the depth at 1 m was too great to be wadeable or too shallow for proper functioning of the D-net. Samples were preserved with 95% ethanol and kept cold through transport to the laboratory, where the ethanol was changed immediately. A fixed count of 500 invertebrates per sample was used, and the total and unsorted portions of the sample were weighed to determine subsampling factors. Invertebrates were identified to three taxonomic resolutions: (1) RBA: a low resolution taxonomic level typical of RBA field protocols (e.g., Ontario Benthos Biomonitoring Network), with the exclusion of lotic taxa (e.g., Simuliidae) and the addition of groups that were abundant in the study region but not typically represented in bioassessment programs (Conchostraca and Chaoboridae); (2) family level; and (3) genus level, the latter two requiring increased time for microscopic examination and because of the amount of taxonomic expertise required.

Data analysis

Connection time (CT) was estimated annually for each lake as the number of days the water level at 10LC002 (extracted from the Environment and Climate Change Canada Real-time Hydrometric Data web site (https://wateroffice.ec.gc.ca/mainmenu/real_time_data_index_e.html) on 21 October 2019) exceeded the elevation measured at the lake outlet. Environmental variables were assessed for normality (Shapiro–Wilk tests) and if necessary square root or log transformations were applied (Table 1) to improve normality prior to analysis, and the major environmental gradients were assessed using principal component analysis (PCA).
Analyses of the invertebrate count data were conducted separately at each taxonomic level to compare the percent variation explained by measured environmental gradients at the three different resolution levels. We first used Moran’s eigenvector maps analysis to assess spatial gradients in the BMI assemblages (Borcard and Legendre 2002). To identify the environmental variables that were important in structuring benthic assemblages we used forward selection to identify the set of variables that maximized the adjusted R2 in a redundancy analysis (RDA). We employed k-means partitioning, a non-hierarchical clustering method, to identify groups of samples with different assemblage types, using the simple structure index (ssi) to determine the best number of groups. Statistical significance of the differences (Bray–Curtis distances) among the groups were assessed with a permutational multivariate analysis of variance as well as pairwise tests using Bonferroni-corrected p values (Martinez Arbizu 2019), and significant taxonomic differences were determined using similarity percentages analysis (Clarke 1993). We then assessed the relevance of these groupings at both taxonomic resolution levels to the environmental variables identified by forward selection using linear discriminant analysis (LDA). The success of the environmental variables at predicting k-means group membership was determined using LDA with jackknifed cross-validation. Analyses were performed in R using package rrcov (Todorov and Filzmoser 2009) for the LDA and package vegan (Oksanen et al. 2019) for all other analyses.

Results

Over five years of sampling a total of approximately 72 000 macroinvertebrates were identified belonging to 135 genera in 65 families. The RBA identification scheme consisted of 22 taxa, ranging in taxonomic level from family to phylum, that are easily identified and tallied in the field. The dominant invertebrate groups were the non-biting midge family Chironomidae (32% of all specimens collected) and freshwater snails (Gastropoda, 20% of all specimens), followed by amphipods, bivalves (8% each), oligochaete and nematode worms (5% each), damselfly larvae (Zygoptera, 5%), clam shrimp (Order Diplostraca, 4%), water mites (3%), water bugs (Hemiptera, 2%) and caddisfly larvae (Trichoptera, 2%). All other insects, apart from the chironomids, made up 14% of the collected specimens, and in addition to groups already mentioned included dragonfly (Anisoptera), mayfly (Ephemeroptera), biting midge (Diptera: Ceratopogonidae), phantom midge (Diptera: Chaoboridae), and horsefly (Tabanidae) larvae, as well as larvae and adults of aquatic beetles (Coleoptera), each totaling <2% of specimens collected.
CT ranged from 0 days in the highest elevation lakes during years of lower spring flood levels to just under 200 days in the connected lakes. Flood conditions during the study period fortuitously captured the range for which historical data are available, with 2013 having high water levels, low levels observed in 2014 and 2016, and average values in 2015 and 2017 (Fig. 2b). PCA of the environmental variables (Table 1) explained 49.9% of the variation on the first two axes, which represented gradients in particulates and nutrients (PC1, explaining 30.4% of the variation) and pH and major ions Ca and K (PC2, explaining 19.5%), with strong contributions of CT and dissolved inorganic carbon (DIC) to both axes. Lake MD4 experienced a sill collapse in the spring of 2014 and subsequent drainage, leading to reduced depth and increased nutrients and particulates in subsequent years. Post-drainage observations at this site were responsible for high maximum values of particulates and phosphorus in Table 1 and were responsible for much of the variation along PC1. When we ran PCA omitting post-drainage observations from MD4 we found the same environmental gradients; however, after removing these samples the pH/ion gradient explained more variation (23.8%) than the particulate/nutrient gradient (16.9%).
Spatial analysis using Moran’s eigenvector maps was not significant for any of the taxonomic levels (RBA: p = 0.899; family: p = 0.642; genus: p = 0.679). This was likely due to the small (∼120 km2) geographic range of the study area relative to the dispersal abilities of many BMI, as the most distant pair of lakes were separated by only ∼12 km, and to the fact that all of the lakes were hydrologically connected at some point during the study period, particularly in the first year of the study (2013; Fig. 2).
We ran k-means partitioning at k = 2 to k = 10 to determine the optimal number of groups for each taxonomic level (Table 2). The ssi was lowest at k = 7 for the RBA level and k = 6 for each of the family and genus levels. Rather than separating into three lake closure groups, analysis of ssi levels suggested that benthic assemblages diverged into multiple clusters: seven groups at the RBA level and six groups at the family and genus levels. To clarify gradients related to connection time, the groups were labelled by their rank mean connection time, with cluster 1 having the lowest CT (1st row of Fig. 3). At the RBA level there were five clusters of unconnected lakes with increasing CT and two groups of lakes with high CT, with cluster 7 exclusively consisting of connected lakes with high CT (Fig. 3a). At the family level (Fig. 3b) there were three clusters of lakes with low CT (family clusters 2–4) and two clusters with high CT (family clusters 5 and 6), with family cluster 1 consisting of MD4 after it drained and a few high-sill sites from 2013, the year with the highest peak flood (Fig. 2b). Family cluster 6 consisted only of lakes that were connected throughout the summer. The genus level k-means clustering was similar to the family level, with variations in group identity for individual samples in clusters 3 and 4 (Fig. 3c).
Table 2.
Table 2. Evaluation of k-means partitioning of the invertebrate samples at different taxonomic resolutions.

Note: Total error sum of squares (SSE) decreases with increasing numbers of clusters, and simple structure index (ssi) is maximized at the number of groups (shown in bold) that best fits the data. RBA, rapid bioassessment approach.

Fig. 3.
Fig. 3. Distribution of important environmental variables within the k-means groups at the (a) rapid bioassessment approach (RBA), (b) family and (c) genus taxonomic levels. Connection time, dissolved inorganic carbon (DIC), SO4, and particulate organic carbon (POC) were identified by forward selection as explaining benthic variation in one or more redundancy analysis models.
Although global analyses of variance of assemblage composition at all taxonomic levels were significant, this was less so for the RBA level (R2 = 0.100, p < 0.002) than the family (R2 = 0.124, p < 0.001) or genus (R2 = 0.123, p < 0.001) levels. Pairwise tests showed all groups to be significantly different at the family and genus levels (Bonferroni adjusted p < 0.03), but this was the case for only about half of the pairs of groups at the RBA level (Table 3). Taxonomic richness was not significantly different between any of the k-means groups at the RBA or family levels (t test, α = 0.05), but generic richness (Fig. 4) was significantly lower in genus cluster 1 (containing the lake that experienced drainage) and genus cluster 3 (containing lakes experiencing thermokarst activity) than in the other groups of High Closure lakes (genus clusters 2 and 4).
Table 3.
Table 3. Pairwise analysis of variance of assemblage composition across k-means groups at different taxonomic levels.

Note: Bonferroni-corrected p values are shown with those with p ≤ 0.03 in bold. RBA, rapid bioassessment approach.

Fig. 4.
Fig. 4. Generic richness among genus-level k-means groups. Significant differences from pairwise t tests (α = 0.05) are indicated by lower case letters.
Similarity percentages analysis of pairwise differences among k-means groups (Table 4) revealed taxa for which significant differences (p < 0.001) in relative abundance contributed to the Bray–Curtis dissimilarities between the groups. As in the global analysis of variance (Table 3) there were fewer significant differences in taxa at the RBA level than at the higher resolutions, particularly among low-CT groups. The RBA groups with higher CT (RBA clusters 5–7) had higher proportions of non-insect BMI, although contrasts with the lowest CT groups were not significant (Table 4). At the family and genus levels, which had similar k-means site classifications, cluster 1 had a depauperate fauna with higher proportions of dorylaimid nematodes and the chironomid Tanytarsus (van der Wulp, 1874) compared with most other clusters. Cluster 3, which had low richness compared with the other low-CT groups also had the greatest number of significant taxa, mainly predatory insects including the tanypode chironomids Ablabesmyia (Johannsen, 1905), Guttipelopia (Fittkau, 1962) and Labrundinia (Fittkau, 1962), the dragonflies Leuchorrhinia (Brittinger, 1850) and Cordulia (Leach, 1815), and the hemipteran Notonecta (Linnaeus, 1758). Cluster 4 also had higher proportions of the dragonfly Aeshna (Fabricius, 1775) and the chironomid Microtendipes (Kieffer, 1915), whereas the other low-CT group (cluster 2) had higher proportions of non-insect grazers including the snails Physa (Draparnaud, 1801) and Fossaria (Westerlund, 1885) and the water mite Limnesia (Koch, 1935). Cluster 5, which contained high-CT lakes that were still usually isolated from the river during the summer, had higher proportions of the chironomids Chironomus (Meigen, 1803) and Einfeldia (Kieffer, 1924), the water beetle Ilybius (Erichson, 1832), and the planorbid snail Gyraulus (Agassiz, 1837). A notable observation in this cluster is the occurrence of the orthocladiine chironomid Propsilocerus (Kieffer, 1923), which despite being common in the Palearctic and in the Nearctic as sub-fossilized remains has only been observed in one other location in North America in a lake in northern British Columbia (Cranston et al. 2011); here, this genus was observed only in the Low Closure lake MD3. Cluster 6, containing connected lakes, had higher proportions of the amphipod Hyalella (Smith, 1874) and large-bodied insects such as Limnephilus (Leach, 1815) and water-boatmen (Corixidae).
Table 4.
Table 4. Taxa identified by similarity percentages (SIMPER) analysis as contributing significantly (p < 0.001) to pairwise differences among k-means groups at different taxonomic levels. Genus level includes nymphs and early instars that could not be identified below the family level.

Note: Only the significant taxa for pairs of sites that had significant overall differences in assemblage composition (Table 3) are shown. Reading down the column for each site group gives the taxa that were more abundant for that group in each comparison. RBA, rapid bioassessment approach.

Forward selection of the environmental variables produced similar sets of variables at each taxonomic level with CT, DIC, pH, and particulate organic carbon (POC) selected in each model, and as the first four variables at the RBA and family levels (Table 5). A larger set of variables was selected explaining the genus level data, with SO4 and CT the first variables selected followed by a number of major ions (Table 5). RDA of the Hellinger-transformed invertebrate counts against the reduced set of environmental variables produced significant models (p < 0.001) at all taxonomic levels (Table 5). Both adjusted R2 and % explained variance increased across the models with increasing taxonomic resolution (RBA: 43.3%, R2 = 0.388; family: 47.0%, R2 = 0.415; genus: 53.1%, R2 = 0.436), although only at the genus level did the RDA model explain >50% of the variation.
Table 5.
Table 5. Redundancy analysis of Hellinger-transformed benthic counts at the rapid bioassessment approach (RBA), family, and genus levels against forward-selected environmental variables.

Note: Only the significant redundancy analysis axes are shown. CT, connection time; DIC, dissolved inorganic carbon; POC, particulate organic carbon; RBA, rapid bioassessment approach; SRP, soluble reactive phosphorus.

There were differences among the taxonomic levels in the relationship between k-means groupings and important environmental variables as determined by forward selection (Fig. 3). DIC increased in the family level groups that were not connected to the river during the summer (family clusters 1–5) as CT increased, but this pattern was not as apparent at the RBA or genus levels. SO4 was very low in genus cluster 3, which mainly consisted of two high elevation lakes experiencing various stages of thermokarst activity (Lakes 520 and 521). POC (as well as particulate organic nitrogen (PON)) was notably higher in RBA cluster 4 and family/genus cluster 1, which contained the lake that drained in 2014 (MD4) and was subsequently more susceptible to wind-driven resuspension of sediments.
When the k-means groups were plotted on the RDA ordinations the higher distinctiveness of groups at the family level was apparent (Fig. 5). The samples from MD4 post-drainage grouped apart at the bottom of the ordination, whereas the remainder of family cluster 1 groups with other samples from High Closure lakes (Fig. 5b). Connected lakes (family cluster 6 and part of family cluster 5) were clearly separated from the unconnected lakes, which fell along a gradient mainly related to DIC, but this separation was less distinct in the RBA and genus levels. In addition, there was some separation of the Low Closure lakes from the High Closures in the family RDA ordination. The RBA level ordination (Fig. 5a) showed both less separation of k-means clusters and greater overlap among a priori closure types, whereas the genus level ordination had distinct k-means clusters but less distinct separation of closure types. Pairwise contrasts among the three a priori closure types were significant at the family level, as were all contrasts between the connected and Low and High closure lakes at all taxonomic levels (Bonferroni-corrected p = 0.003) but contrasts between the Low and High closure groups were not significant at the RBA level (p = 0.120) and only slightly significant at the genus level (p = 0.006).
Fig. 5.
Fig. 5. Redundancy analysis biplots (scaling 1) showing sites after forward selection of environmental (explanatory) variables at (a) rapid bioassessment approach (RBA) taxonomic level, (b) family level, and (c) genus level. Symbol shape and shading corresponds to the k-means partitions for each taxonomic level. The ellipses enclose all members of each of the a priori lake closure groups: solid line = High Closure, dashed line = Low Closure, dotted line = No Closure. Cond, conductivity; CT, connection time; DIC, dissolved inorganic carbon; POC, particulate organic carbon.
There were similarities among the taxonomic levels in the species responses to environmental gradients (Fig. 6). In each analysis, relative abundance of the clam shrimp Lynceus brachyurus (Müller, 1776) was directly associated with CT. Chironomids (Tanytarsini: Tanytarsus in the family and genus analyses) were negatively associated with CT and Hyalella (or Amphipoda) were positively associated with CT at all three levels. A secondary gradient was present in the RBA and family analyses involving two or more of the variables DIC, pH, Si, and POC that was associated with a taxonomic gradient from snails (Family: Planorbiidae) and damselfly larvae (Family: Coenagrionidae) to Chaoborus (Lichtenstein, 1800), dorylaimid nematodes, and corixid nymphs.
Fig. 6.
Fig. 6. Redundancy analysis biplots (scaling 2) showing responses of (a) rapid bioassessment approach (RBA) taxa, (b) families and (c) genera to gradients in forward-selected environmental variables. For clarity, only taxa with scores above a threshold on either redundancy analysis (RDA) axis were plotted; these were >0.05 for the RBA level and >0.1 for the family and genus level ordinations. Cond, conductivity; CT, connection time; DIC, dissolved inorganic carbon; POC, particulate organic carbon.
Prior to linear discriminant analysis the k-means groups were tested for multivariate homogeneity of within-group covariance (RBA: p = 0.015; family: p = 0.36; fenus: p = 0.254) and differences of means using Wilk’s λ test (RBA: λ = 0.024, p < 0.001; family: λ = 0.0049, p < 0.001; genus: λ = 0.0041, p < 0.001). LDA with jack-knife cross-validation (Table 6) indicated that congruence between groupings based on environmental variables and BMI assemblages increased with increasing taxonomic resolution. The RBA level had the lowest success rate (51% of sites classified correctly) and the greatest number of different errors (non-zero results in the non-diagonal cells of Table 6a). Jack-knifed classification success at the family level (Table 6b) was considerably higher (76%) and classified most of the sites in each group correctly. The genus level had the highest success rate in the LDA (84%) and classified some groups (genus clusters 2 and 4) with 100% accuracy.
Table 6.
Table 6. Contingency tables showing classification (%) of sites based on environmental variables compared with k-means classification of benthic macroinvertebrate assemblages at (a) rapid bioassessment approach (RBA), (b) family and (c) genus taxonomic levels.

Note: Diagonals in bold are the classification success rates (%) for each k-means group.

Discussion

Comparison of taxonomic level

The required taxonomic sufficiency of biomonitoring programs is a contentious issue. The increased information afforded by high-resolution identification to genus/species must be balanced by consideration of the higher cost in personnel training and expertise, money, and time relative to lower resolution identification (Jones 2008). Many studies have found acceptably congruent results between family-level and genus/species-level identifications in representing assemblage structure (Törnblom et al. 2011; Kallimanis et al. 2012; Landeiro et al. 2012; Carew et al. 2018) and assemblage–environment relationships (Chessman et al. 2007; Schmera and Erős 2011; Vilmi et al. 2016). In many cases the family level is sufficient to establish the overall or most important environmental relationships, but with reduced statistical power (Hawkins et al. 2000; Monk et al. 2012) or loss of ability to detect subtler environmental changes or relationships (Greffard et al. 2011; Forcino et al. 2012; Jiang et al. 2013). Key factors determining the success of broad taxonomic resolutions are the number and ecological diversity of species being aggregated into larger groups (Bennett et al. 2014; dos Santos Ribas and Padial 2015; Lu et al. 2016) and assemblage richness at the broader taxonomic resolution (Hawkins et al. 2000; Milošević et al. 2014).
Jones (2008) identified a number of analytical considerations pertaining to the multivariate statistical methods that we used. First, agglomerating taxa lowers the number of variables available for analysis, in our case from 135 genera to 64 families to 22 RBA taxa. Agglomeration therefore decreases the total variance of the multivariate data (Table 5). Second, as higher levels of agglomeration are used the informative value of presence/absence data declines. More than half of our RBA taxa (compared with approximately 1/6 of our families and only 9% of our genera) were present in every sample (Fig. 7a). If we were to agglomerate taxa higher than the RBA level (e.g., to class), presence–absence data would produce the trivial result of every taxon being present in every sample. Presence–absence RDA at every level produced much lower adjusted R2 values (RBA: 0.293, family: 0.332, genus: 0.292). This is unfortunate from a rapid bioassessment perspective as collection of presence–absence data is more time-efficient than generating count data in a field survey. Third, agglomeration decreases the number of potential distinct sources of information available to elucidate species–environment relationships in the canonical analysis and to make ecological interpretations. Like many freshwater environments, our study lakes were dominated by the family Chironomidae (here considered an RBA taxon, divided into subfamilies/tribes in our family analysis). Chironomids are a species-rich and ecologically diverse family, so by agglomerating them into one variable, information is lost compared with the more equitable division of chironomid subfamilies/tribes in the family analysis (Fig. 7b). This is also true of less numerically dominant, but taxonomically, diverse RBA groups (e.g., Trichoptera, Coleoptera). This loss of information may decrease the ability to detect subtle environmental changes (Jones 2008; Greffard et al. 2011) and may explain why the family and genus levels were able to distinguish between connected and closed lakes (Fig. 5) and had higher classification success in the LDA (Table 6).
Fig. 7.
Fig. 7. (a) Occurrence frequencies of rapid bioassessment approach RBA taxa, families, and genera. (b) Relative abundance of Chironomidae (RBA analysis, shaded box) and chironomid subfamilies/tribes (family analysis, unshaded boxes).

Relationship to environmental gradients

We expected CT to be an important variable structuring the benthic assemblages, and this was largely confirmed in our analyses. At each taxonomic level, k-means classification produced groups that differed significantly in CT (Fig. 3), and CT was among the first variables included in the forward selection RDA models (Table 5) and explained large amounts of benthic community variation in ordinations at all taxonomic levels (Fig. 5). The observation that CT had the highest loading on the first axis of the RBA-level RDA supports the importance of CT in structuring the benthic assemblages. However, at the RBA resolution there was little apparent relationship between assemblage types defined by k-means partitioning and limnological variables or separation among either a priori closure types or k-means groups in the ordination (Fig. 5a). At the family level the No Closure lakes occupied a distinct ordination space (top left quadrant of Fig. 5b), whereas clusters 2–5 formed a gradient along RDA 2 with little overlap in the clusters compared with the RBA analysis in which there was a great deal of overlap among the low-CT k-means groups. At the genus level a distinct set of explanatory variables were selected through RDA analysis (Table 5), including SO4, which explained more variation than CT, and there was more overlap among closure types (Fig. 5b) than in the family ordination, although more total variation was explained at the genus level (Table 5). SO4 has multiple sources in Mackenzie Delta lakes (Lesack et al. 1998), either being introduced by the river via flooding or distributary channels in connected lakes or released from the sediment in very shallow lakes, such as MD4 (cluster 1) in this study.
Despite the importance of CT as an explanatory variable, much of the variation among low-CT assemblages at the family level was explained by a secondary gradient that reflected lake productivity. Previous work in the Mackenzie Delta has established a gradient of lake primary productivity between phytoplankton-dominated production in turbid, connected lakes and macrophyte-dominated production — along with associated epiphyton — in relatively clear, unconnected lakes (Squires et al. 2009). In connected lakes, river inputs provide suspended solids that limit the growth of macrophytes and benthic algae, but also nutrients that stimulate the growth of phytoplankton closer to the surface (Squires and Lesack 2002). Lakes with short connection times or that infrequently flood have lower river-derived inorganic turbidity (Squires and Lesack 2003) and nutrients, and primary production is dominated by macrophytes, which derive their nutritional requirements from nutrient-rich sediments and reach very high biomass (>2000 g/m2) in the lakes with greatest water transparency (Squires et al. 2002). During the summer macrophyte photosynthesis can draw down dissolved inorganic carbon and cause elevated pH levels (Hesslein et al. 1991; Tank et al. 2009a, 2009b) and high macrophyte biomass limits the light available to phytoplankton and epipelic algae via shading, while providing surfaces for growth of epiphyton (Squires et al. 2009). This turbidity-driven productivity gradient was reflected in RDA 2 of the family-level ordination (Fig. 5b), with a gradient primarily from high-DIC (low primary productivity) lakes to low DIC (highly productive lakes). Evidence from stable isotopes suggest that at least some primary consumers (zooplankton and molluscs) derive more carbon from phytoplankton in turbid, connected lakes and rely on benthic algae (epipelon and epiphyton) in unconnected lakes with large macrophyte communities (Hecky and Hesslein 1995). Bacteria are likely to be a major food source for many benthic grazers (e.g., most Chironomidae), and previous work has demonstrated bacterial production and activity to vary along the described productivity gradient although the relationship is more complex and less fully understood. Bacterioplankton abundance and production is higher in more isolated lakes than connected lakes (Spears and Lesack 2006), possibly due to greater supplies of autochthonous DOC from macrophytes (Tank et al. 2011), but are limited by alkalinity-related stress when photosynthetic rates are high enough to drive up pH (Tank et al. 2009b). Macrophyte-mediated differences in DOC quantity and quality are also important drivers of the activity of methanogenic microbial communities (Cunada et al. 2018), which can also be important food sources for benthic invertebrates (e.g., Yasuno et al. 2013).
Given the above explanation for the major environmental gradients, we interpret the constrained ordination of benthic families and genera as follows: No Closure lakes (cluster 6) have distinct assemblages, including higher proportions of crustaceans (Lynceus, Hyalella) and large-bodied insects (Corixidae, Limnephilus), from lakes that become disconnected during the summer or flood less than once per year, which have greater proportions of insects, particularly Chironomidae and Odonata (Anisoptera and Zygoptera), capable of aerial dispersal. When the lakes have sills high enough to separate the lake from the river channels, assemblages are structured along a gradient of lake primary productivity, which as described above is strongly related to CT. Among these, the more productive lakes (cluster 2) have greater proportions of grazers (e.g., Gyraulus) that may be supported by abundant epiphyton (Squires et al. 2009), whereas more turbid water in the lower-sill lakes (cluster 5) limits macrophyte productivity and dissolved O2, leading to a distinct community type dominated by Chironomus spp. Lakes that are cut off from the river for >1 year have distinct communities depending on the ecological trajectory of the lake during the period between inundations (Lesack and Marsh 2010). Shallow lakes subject to evaporative losses lack many taxa present elsewhere in the study area (cluster 1). Deeper high-sill lakes (cluster 3) may be disconnected for multiple years and are often subject to thermokarst activity, and are distinguished from similarly low-CT lakes in having greater proportions of invertebrate predators.

Implications for biomonitoring and future ecological changes

Based on our comparison of the performance of the different taxonomic levels, we conclude that a biomonitoring effort based on at least the family level is sufficient to detect changes in lakes in the Mackenzie Delta in response to changes in hydrological connectivity. Although CT was an important structuring variable, a minimal RBA approach was less able to differentiate between connected and isolated lakes and is therefore less likely to be useful in detecting subtler effects due to changes in primary productivity or isolation of High Closure lakes. Our family-level results were likely also improved by the separation of subfamilies/tribes of the dominant family Chironomidae. The family level analysis also showed closer correspondence between k-means groups defined by the assemblage and group membership predicted by LDA on the environmental variables (Table 6). Identifying BMI further to the genus level resulted in identical site classification to the family analysis, and moderate improvements in the RDA model and LDA performance. Conversely, differences in richness were only apparent at the genus level, as were additional ecological details such as increased proportions of predators in high-sill thermokarst lakes and the presence of an extremely uncommon chironomid in a small Low Closure lake. Therefore, it seems that detailed ecological or biodiversity studies of BMI communities should identify specimens to the genus or species level, whereas family identification may be sufficient for routine monitoring.
Our study is the first to examine differences in invertebrate communities in lakes along the connectivity gradient in the Delta. The strong relationship between benthic assemblages and connection time in our study indicates that the former will be a useful indicator of the changes that are expected to occur to the hydrology and ecology of the Delta under climate warming scenarios. Peak water levels in the spring determine the connectivity and degree of flushing of the ∼45 000 lakes in the Delta (Marsh and Hey 1989; Emmerton et al. 2007) and the peak levels are determined by ice-breakup processes to a far greater degree than increases in discharge (Rood et al. 2017), both of which are associated with climatic changes (Lesack et al. 2013). Mechanical breakup of the ice cover that causes ice-jam flooding is temperature-sensitive (Beltaos 2013), and declines in the date of ice breakup (Marsh et al. 2002) and the lag time between onset of spring discharge and ice breakup (Lesack et al. 2013) have been attributed to increases in local spring air temperature (Lesack et al. 2014; Wang et al. 2017). Declines in ice-jam-driven flooding likely will lead to decreased connection times of high elevation lakes (Lesack and Marsh 2007), which may dry up if not flooded for many years (Emmerton et al. 2007). In contrast, low elevation lakes are likely to experience increased connection times due to rising sea levels in the northern Delta and possibly longer ice-free seasons (Emmerton et al. 2007; Lesack and Marsh 2007). Lesack and Marsh (2010) proposed that a consequence of these changes would be high elevation lakes becoming increasingly ecologically distinct from each other while also declining in abundance, whereas abundance and ecological similarity of low elevation or connected lakes is expected to increase, with a net result of a loss of ecological diversity (Lesack and Marsh 2010) and a decrease in storage and transformation of flood waters before delivery to the Arctic Ocean (Emmerton et al. 2007, 2008). Because we found that BMI responded to variation in CT, benthic assemblages should be useful as rapid indicators of ecological change due to varying flood levels, as well as long-term indicators of the ecological changes that have been predicted including greater ecological diversity in high elevation lakes (Lesack and Marsh 2010).
Our results indicate that CT is of fundamental importance to benthic community structure even when the community is described at a coarse taxonomic level, and that BMI at the family level or lower are useful for monitoring changes to lake ecosystems related to altered flooding hydrology (e.g., gradients in primary productivity in Low Closure lakes). Our family-level ordination suggests that with increasing time between floods, BMI communities in High Closure lakes (which are somewhat distinct from Low Closure lakes in Fig. 5) will become more distinct and occupy unique ordination spaces from lakes that regularly flood, as lack of dispersal routes limit opportunities for colonization and changes to habitat quality alter the food resources and niches available. Increases in CT in the Low Closure lakes would increase dispersal opportunities for BMI and increase similarity in limnological characteristics among these lakes, so BMI communities in these lakes may become more homogeneous and similar to the No Closure lakes, which formed a distinct community type in our analysis. The precise effect of the predicted changes in CT on lake ecology is difficult to predict given current knowledge because as lakes become more isolated they may follow unique trajectories based on the relative importance of evaporation and thermokarst, as well as variation in conditions prior to isolation (Lesack and Marsh 2010). Given the environmental complexity of the Mackenzie Delta system, the use of an integrative bioindicator such as BMI communities would be of great use in monitoring the effects of current and future climate change.

Acknowledgements

This work was supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery and NSERC Discovery Northern Research Supplement grants to R.Q. and Northern Scientific Training Program support to R.W.S. Logistical and technical support was provided by the Western Arctic Research Centre, Aurora Research Institute, in Inuvik, Northwest Territories, and we express thanks to Jolie Gareis, William Hurst and Edwin Amos for their valuable advice and assistance. We thank Andrew Medeiros, Christopher Luszczek, Sara Masood, Frankie Talarico, Cait Carew, and Dmitri Perlov from York University for assistance in the field. We are also grateful to Professor Lance Lesack from Simon Fraser University for providing valuable advice on calculating connection times. We also thank two anonymous reviewers for their valuable suggestions on a previous version of this manuscript. This research was conducted as a part of Scientific Research Licenses 15272, 15403, 15652, 15811, and 16149 from the Aurora Research Institute.

References

ACIA. 2005. Arctic climate impact assessment. Cambridge University Press, Cambridge, UK. 1042 pp.
Beltaos S. 2013. Hydrodynamic and climatic drivers of ice breakup in the lower Mackenzie River. Cold Reg. Sci. Technol. 95: 39–52.
Bennett J.R., Sisson D.R., Smol J.P., Cumming B.F., Possingham H.P., Buckley Y.M., and Cadotte M. 2014. Optimizing taxonomic resolution and sampling effort to design cost-effective ecological models for environmental assessment. J. Appl. Ecol. 51(6): 1722–1732.
Borcard D. and Legendre P. 2002. All-scale spatial analysis of ecological data by means of principal coordinates of neighbor matrices. Ecol. Modell. 153(1–2): 51–68.
Borisko J.P., Kilgour B.W., Stanfield L.W., and Jones F.C. 2007. An evaluation of rapid bioassessment protocols for stream benthic invertebrates in southern Ontario, Canada. Water Qual. Res. J. 42(3): 184–193.
Bouchard F., MacDonald L.A., Turner K.W., Thienpont J.R., Medeiros A.S., Biskaborn B.K., et al. 2017. Paleolimnology of thermokarst lakes: a window into permafrost landscape evolution. Arct. Sci. 3(2): 91–117.
Burn C.R. and Kokelj S.V. 2009. The environment and permafrost of the Mackenzie Delta area. Permafr. Periglac. Process. 20(2): 83–105.
Carew M.E., Kellar C.R., Pettigrove V.J., and Hoffman A.A. 2018. Can high-throughput sequencing detect macroinvertebrate diversity for routine monitoring of an urban river? Ecol. Indic. 85: 440–450.
Chessman B., Williams S., and Besley C. 2007. Bioassessment of streams with macroinvertebrates: effect of sampled habitat and taxonomic resolution. J. North Am. Benthol. Soc. 26(3): 546–565.
Clarke K.R. 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18(1): 117–143.
Cranston P.S., Barley E., Langley G.E., Dieffenbacher-Krall A., Longmuir A., and Zloty J. 2011. Propsilocerus Kieffer (Diptera: Chironomidae) from the Nearctic. Aquat. Insects, 33(4): 343–350.
Culp J.M., Lento J., Curry R.A., Luiker E., and Halliwell D. 2019. Arctic biodiversity of stream macroinvertebrates declines in response to latitudinal change in the abiotic template. Freshwater Sci. 38(3): 465–479.
Cunada C.L., Lesack L.F.W., and Tank S.E. 2018. Seasonal dynamics of dissolved methane in lakes of the Mackenzie Delta and the role of carbon substrate quality. J. Geophys. Res.: Biogeosci. 123(2): 591–609.
dos Santos Ribas L.G. and Padial A.A. 2015. The use of coarser data is an effective strategy for biological assessments. Hydrobiologia, 747(1): 83–95.
Emmerton C.A., Lesack L.F.W., and Marsh P. 2007. Lake abundance, potential water storage, and habitat distribution in the Mackenzie River Delta, western Canadian Arctic. Water Resour. Res. 43: W05419.
Emmerton C.A., Lesack L.F.W., and Vincent W.F. 2008. Mackenzie River nutrient delivery to the Arctic Ocean and effects of the Mackenzie Delta during open water conditions. Global Biogeochem. Cycles, 22: GB1024.
Environment Canada. 2017. SOP 01–090, Standard operating procedure for the analysis of carbon and nitrogen in suspended particulate matter and sediment in natural water via combustion procedure. National Laboratory for Environmental Testing, Burlington, Ont., Canada.
Forcino F.L., Stafford E.S., and Leighton L.R. 2012. Perception of paleocommunities at different taxonomic levels: how low must you go? Palaeogeogr. Palaeoclimatol., Palaeoecol. 365–366: 48–56.
Funk A., Trauner D., Reckendorfer W., and Hein T. 2017. The Benthic Invertebrates Floodplain Index — extending the assessment approach. Ecol. Indic. 79: 303–309.
Gallardo B., García M., Cabezas Á., González E., González M., Ciancarelli C., and Comín F.A. 2008. Macroinvertebrate patterns along environmental gradients and hydrological connectivity within a regulated river-floodplain. Aquat. Sci. 70(3): 248–258.
Gallardo B., Dolédec S., Paillex A., Arscott D.B., Sheldon F., Zilli F., et al. 2014. Response of benthic macroinvertebrates to gradients in hydrological connectivity: a comparison of temperate, subtropical, Mediterranean and semiarid river floodplains. Freshwater Biol. 59(3): 630–648.
González-Bergonzoni I., Landkildehus F., Meerhoff M., Lauridsen T.L., Özkan K., Davidson T.A., et al. 2014. Fish determine macroinvertebrate food webs and assemblage structure in Greenland subarctic streams. Freshwater Biol. 59(9): 1830–1842.
Government of Northwest Territories. 2018. NWT Cumulative Impact Monitoring Program 2017/18 Annual Report. Available from https://www.enr.gov.nt.ca/en/services/cumulative-impact-monitoring-program-cimp/resources-nwt-cimp [accessed 11 October 2019].
Greffard M.-H., Saulnier-Talbot É., and Gregory-Eaves I. 2011. A comparative analysis of fine versus coarse taxonomic resolution in benthic chironomid community analyses. Ecol. Indic. 11(6): 1541–1551.
Hawkins C.P., Norris R.H., Hogue J.N., and Feminella J.W. 2000. Development and evaluation of predictive models for measuring the biological integrity of streams. Ecol. Appl. 10(5): 1456–1477.
Hecky R.E. and Hesslein R.H. 1995. Contributions of benthic algae to lake food webs as revealed by stable isotope analysis. J. North Am. Benthol. Soc. 14(4): 631–653.
Hesslein, R.H., Rudd, J.W.M., Kelly, C., Ramlal, P., and Hallard, K.A. 1991. Carbon dioxide pressure in surface waters of Canadian lakes. In Air-water mass transfer: selected papers from the Second International Symposium on Gas Transfer at Water Surfaces. Edited by S.C. Wilhelms and J.S. Gulliver. American Society of Civil Engineers, New York, N.Y., USA. pp. 413–431.
Holmes, R.M., Coe, M.T., Fiske, G.J., Gurtovaya, T., McClelland, J.W., Shklomanov, A.I., et al. 2012. Climate change impacts on the hydrology and biogeochemistry of Arctic rivers. In Climatic change and global warming of inland waters: impacts and mitigation for ecosystems and societies. Edited by C.R. Goldman, M. Kumagai, and R.D. Robarts. John Wiley & Sons, Ltd., Chichester, UK. pp. 1–26.
Huttunen K.-L., Mykrä H., Paavola R., and Muotka T. 2018. Estimates of benthic invertebrate community variability and its environmental determinants differ between snapshot and trajectory designs. Freshwater Sci. 37(4): 769–779.
Jiang X., Xiong J., Song Z., Morse J.C., Jones F.C., and Xie Z. 2013. Is coarse taxonomy sufficient for detecting macroinvertebrate patterns in floodplain lakes? Ecol. Indic. 27: 48–55.
Jones F.C. 2008. Taxonomic sufficiency: the influence of taxonomic resolution on freshwater bioassessments using benthic macroinvertebrates. Environ. Rev. 16: 45–69.
Junk, W.J., Bayley, P.B., and Sparks, R.E. 1989. The flood pulse concept in river-floodplain systems. In Proceedings of the International Large River Symposium (LARS). Edited by D.P. Dodge. Canadian Special Publication of Fisheries and Aquatic Sciences 106. pp. 110–127.
Kallimanis A.S., Mazaris A.D., Tsakanikas D., Dimopoulos P., Pantis J.D., and Sgardelis S.P. 2012. Efficient biodiversity monitoring: which taxonomic level to study? Ecol. Indic. 15(1): 100–104.
Kendrick M.R., Huryn A.D., Bowden W.B., Deegan L.A., Findlay R.H., Hershey A.E., et al. 2018. Linking permafrost thaw to shifting biogeochemistry and food web resources in an arctic river. Glob. Change Biol. 24(12): 5738–5750.
Kharouba H.M., Ehrlén J., Gelman A., Bolmgren K., Allen J.M., Travers S.E., and Wolkovich E.M. 2018. Global shifts in the phenological synchrony of species interactions over recent decades. Proc. Natl. Acad. Sci. U.S.A. 115(20): 5211–5216.
Klobucar S.L., Gaeta J.W., and Budy P. 2018. A changing menu in a changing climate: using experimental and long-term data to predict invertebrate prey biomass and availability in lakes of arctic Alaska. Freshwater Biol. 63(11): 1352–1364.
Krogh S.A. and Pomeroy J.W. 2018. Recent changes to the hydrological cycle of an Arctic basin at the tundra–taiga transition. Hydrol. Earth Syst. Sci. 22(7): 3993–4014.
Lafrenière M.J. and Lamoureux S.F. 2019. Effects of changing permafrost conditions on hydrological processes and fluvial fluxes. Earth Sci. Rev. 191: 212–223.
Landeiro V.L., Bini L.M., Costa F.R.C., Franklin E., Nogueira A., de Souza J.L.P., et al. 2012. How far can we go in simplifying biomonitoring assessments? An integrated analysis of taxonomic surrogacy, taxonomic sufficiency and numerical resolution in a megadiverse region. Ecol. Indic. 23: 366–373.
Laske S.M., Haynes T.B., Rosenberger A.E., Koch J.C., Wipfli M.S., Whitman M., and Zimmerman C.E. 2016. Surface water connectivity drives richness and composition of Arctic lake fish assemblages. Freshwater Biol. 61(7): 1090–1104.
Lesack L.F.W. and Marsh P. 2007. Lengthening plus shortening of river-to-lake connection times in the Mackenzie River Delta respectively via two global change mechanisms along the arctic coast. Geophys. Res. Lett. 34: L23404.
Lesack L.F.W. and Marsh P. 2010. River-to-lake connectivities, water renewal, and aquatic habitat diversity in the Mackenzie River Delta. Water Resour. Res. 46: W12504.
Lesack L.F.W., Marsh P., and Hecky R.E. 1998. Spatial and temporal dynamics of major solute chemistry among Mackenzie Delta lakes. Limnol. Oceanogr. 43(7): 1530–1543.
Lesack L.F.W., Marsh P., Hicks F.E., and Forbes D.L. 2013. Timing, duration, and magnitude of peak annual water-levels during ice breakup in the Mackenzie Delta and the role of river discharge. Water Resour. Res. 49(12): 8234–8249.
Lesack L.F.W., Marsh P., Hicks F.E., and Forbes D.L. 2014. Local spring warming drives earlier river-ice breakup in a large Arctic delta. Geophys. Res. Lett. 41(5): 1560–1567.
Lininger K.B. and Wohl E. 2019. Floodplain dynamics in North American permafrost regions under a warming climate and implications for organic carbon stocks: a review and synthesis. Earth Sci. Rev. 193: 24–44.
Lu H.-P., Yeh Y.-C., Sastri A.R., Shiah F.-K., Gong G.-C., and Hsieh C.-H. 2016. Evaluating community–environment relationships along fine to broad taxonomic resolutions reveals evolutionary forces underlying community assembly. ISME J. 10(12): 2867–2878.
MacDonald L.A., Wolfe B.B., Turner K.W., Anderson L., Arp C.D., Birks S.J., et al. 2017. A synthesis of thermokarst lake water balance in high-latitude regions of North America from isotope tracers. Arct. Sci. 3(2): 118–149.
Mackay, J.R. 1963. The Mackenzie Delta area, N.W.T. Geographical Branch, Memoir 8. Department of Energy, Mines and Resources, Geological Survey of Canada, Ottawa, Ont., Canada. 202 pp.
Mackay J.R. 1995. Active layer changes (1968 to 1993) following the forest-tundra fire near Inuvik, N.W.T., Canada. Arct. Antarct. Alp. Res. 27(4): 323–336.
Marsh, P., and Hey, M. 1988. Mackenzie River water levels and the flooding of delta lakes. Contribution 88013. National Hydrology Research Institute, Environment Canada, Saskatoon, Sask., Canada.
Marsh P. and Hey M. 1989. The flooding hydrology of Mackenzie Delta lakes near Inuvik, N.W.T., Canada. Arctic, 42: 41–49.
Marsh P. and Lesack L.F.W. 1996. The hydrologic regime of perched lakes in the Mackenzie Delta: potential responses to climate change. Limnol. Oceanogr. 41: 849–856.
Marsh P., Onclin C., and Neumann N. 2002. Water and energy fluxes in the lower Mackenzie valley, 1994/95. Atmos.-Ocean, 40(2): 245–256.
Martinez Arbizu, P. 2019. pairwiseAdonis: pairwise multilevel comparison using adonis. R package version 0.3.
Medeiros A.S., Luszczek C.E., Shirley J., and Quinlan R. 2011. Benthic biomonitoring in Arctic tundra streams: a community-based approach in Iqaluit, Nunavut, Canada. Arctic, 64(1): 59–72.
Milošević D., Stojković M., Čerba D., Petrović A., Paunović M., and Simić V. 2014. Different aggregation approaches in the chironomid community and the threshold of acceptable information loss. Hydrobiologia, 727(1): 35–50.
Monk W.A., Wood P.J., Hannah D.M., Extence C.A., Chadd R.P., and Dunbar M.J. 2012. How does macroinvertebrate taxonomic resolution influence ecohydrological relationships in riverine ecosystems Ecohydrology, 5(1): 36–45.
Oksanen, J., Blanchet, F.G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., et al. 2019. vegan: community ecology package. R package version 2.5-4. Available from https://CRAN.R-project.org/package=vegan.
Palmer M.J., Burn C.R., Kokelj S.V., and Allard M. 2012. Factors influencing permafrost temperatures across tree line in the uplands east of the Mackenzie Delta, 2004–2010. Can. J. Earth Sci. 49(8): 877–894.
Petsch D.K., Pinha G.D., and Takeda A.M. 2017. Dispersal mode and flooding regime as drivers of benthic metacommunity structure in a Neotropical floodplain. Hydrobiologia, 788(1): 131–141.
Rood S.B., Kaluthota S., Philipsen L.J., Rood N.J., and Zanewich K.P. 2017. Increasing discharge from the Mackenzie River system to the Arctic Ocean: Mackenzie River System. Hydrol. Process. 31(1): 150–160.
Rosenberg, D.M., and Resh, V.H. (Editors). 1993. Freshwater biomonitoring and benthic macroinvertebrates. Chapman & Hall, New York, N.Y., USA.
Rosenberg, D.M., and Snow, N.B. 1975. Ecological studies of aquatic organisms in the Mackenzie and Porcupine River drainages in relation to sedimentation. Fisheries and Marine Service Technical Report No. 547. Freshwater Institute, Winnipeg, Man., Canada. 86 pp.
Rouse W.R., Douglas M.S.V., Hecky R.E., Hershey A.E., Kling G.W., Lesack L.F.W., et al. 1997. Effects of climate change on the freshwaters of arctic and subarctic North America. Hydrol. Process. 11(8): 873–902.
Ruetz C., Newman R., and Vondracek B. 2002. Top-down control in a detritus-based food web: fish, shredders, and leaf breakdown. Oecologia, 132(2): 307–315.
Scarabotti P.A., Demonte L.D., and Pouilly M. 2017. Climatic seasonality, hydrological variability, and geomorphology shape fish assemblage structure in a subtropical floodplain. Freshwater Sci. 36(3): 653–668.
Schmera D. and Erős T. 2011. The role of sampling effort, taxonomical resolution and abundance weight in multivariate comparison of stream dwelling caddisfly assemblages collected from riffle and pool habitats. Ecol. Indic. 11(2): 230–239.
Spears B.M. and Lesack L.F.W. 2006. Bacterioplankton production, abundance, and nutrient limitation among lakes of the Mackenzie Delta (western Canadian arctic). Can. J. Fish. Aquat. Sci. 63(4): 845–857.
Squires M.M. and Lesack L.F.W. 2002. Water transparency and nutrients as controls on phytoplankton along a flood-frequency gradient among lakes of the Mackenzie Delta, western Canadian Arctic. Can. J. Fish. Aquat. Sci. 59(8): 1339–1349.
Squires M.M. and Lesack L.F.W. 2003. Spatial and temporal patterns of light attenuation among lakes of the Mackenzie Delta. Freshwater Biol. 48(1): 1–20.
Squires M.M., Lesack L.F.W., and Huebert D. 2002. The influence of water transparency on the distribution and abundance of macrophytes among lakes of the Mackenzie Delta, Western Canadian Arctic. Freshwater Biol. 47(11): 2123–2135.
Squires M.M., Lesack L.F.W., Hecky R.E., Guildford S.J., Ramlal P., and Higgins S.N. 2009. Primary production and carbon dioxide metabolic balance of a lake-rich arctic river floodplain: partitioning of phytoplankton, epipelon, macrophyte, and epiphyton production among lakes on the Mackenzie Delta. Ecosystems, 12(5): 853–872.
Stewart R.E., Crawford R.W., Leighton H.G., Marsh P., Strong G.S., Moore G.W.K., et al. 1998. The Mackenzie GEWEX Study: the water and energy cycles of a major North American river basin. Bull. Am. Meteorol. Soc. 79(12): 2665–2683.
Tank S.E., Lesack L.F.W., and Hesslein R.H. 2009a. Northern delta lakes as summertime CO2 absorbers within the arctic landscape. Ecosystems, 12(1): 144–157.
Tank S.E., Lesack L.F.W., and McQueen D.J. 2009b. Elevated pH regulates bacterial carbon cycling in lakes with high photosynthetic activity. Ecology, 90(7): 1910–1922.
Tank S.E., Lesack L.F.W., Gareis J.A.L., Osburn C.L., and Hesslein R.H. 2011. Multiple tracers demonstrate distinct sources of dissolved organic matter to lakes of the Mackenzie Delta, western Canadian Arctic. Limnol. Oceanogr. 56(4): 1297–1309.
Thornhill I., Loiselle S., Clymans W., and van Noordwijk C.G.E. 2019. How citizen scientists can enrich freshwater science as contributors, collaborators, and co-creators. Freshwater Sci. 38(2): 231–235.
Tockner K., Malard F., and Ward J.V. 2000. An extension of the flood pulse concept. Hydrol. Process. 14: 2861–2883.
Todorov V. and Filzmoser P. 2009. An object-oriented framework for robust multivariate analysis. J. Stat. Softw. 32(3): 1–47.
Törnblom J., Roberge J.-M., and Angelstam P. 2011. Rapid assessment of headwater stream macroinvertebrate diversity: an evaluation of surrogates across a land-use gradient. Fundam. Appl. Limnol. 178(4): 287–300.
Vilmi A., Karjalainen S.M., Nokela T., Tolonen K., and Heino J. 2016. Unravelling the drivers of aquatic communities using disparate organismal groups and different taxonomic levels. Ecol. Indic. 60: 108–118.
Vonk J.E., Tank S.E., Bowden W.B., Laurion I., Vincent W.F., Alekseychik P., et al. 2015. Reviews and syntheses: effects of permafrost thaw on Arctic aquatic ecosystems. Biogeosciences, 12(23): 7129–7167.
Wang S., Zhou F., and Russell H. 2017. Estimating snow mass and peak river flows for the Mackenzie River basin using GRACE satellite observations. Remote Sens. 9(3): 256.
Wiens, A.P., Rosenberg, D.M., and Snow, N.B. 1975. Species list of aquatic plants and animals collected from the Mackenzie and Porcupine River watersheds from 1971 to 1973. Canada Department of the Environment Fisheries and Marine Service Technical Report No. 557. Freshwater Institute, Winnipeg, Man., Canada. 39 pp.
Yasuno N., Shikano S., Shimada T., Shindo K., and Kikuchi E. 2013. Comparison of the exploitation of methane-derived carbon by tubicolous and non-tubicolous chironomid larvae in a temperate eutrophic lake. Limnology, 14(3): 239–246.

Information & Authors

Information

Published In

cover image Arctic Science
Arctic Science
Volume 6Number 4December 2020
Pages: 463 - 487

History

Received: 21 October 2019
Accepted: 29 May 2020
Version of record online: 6 August 2020

Key Words

  1. benthic invertebrates
  2. biomonitoring
  3. Mackenzie Delta
  4. floodplain lakes
  5. limnology

Mots-clés

  1. invertébrés benthiques
  2. biosurveillance
  3. delta du Mackenzie
  4. lacs des plaines inondables
  5. limnologie

Authors

Affiliations

Ryan W. Scott [email protected]
Department of Biology, York University, 4700 Keele Street, Toronto, ON M3J 2V1, Canada.
Suzanne E. Tank
Department of Biological Sciences, University of Alberta, Edmonton, TG6 2E9, Canada.
Xiaowa Wang
Environment and Climate Change Canada, 867 Lakeshore Road, Burlington, ON L7R 4A6, Canada.
Roberto Quinlan
Department of Biology, York University, 4700 Keele Street, Toronto, ON M3J 2V1, Canada.

Notes

Copyright remains with the author(s) or their institution(s). This work is licensed under a Creative Attribution 4.0 International License (CC BY 4.0) http://creativecommons.org/licenses/by/4.0/deed.en_GB, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

Metrics & Citations

Metrics

Other Metrics

Citations

Cite As

Export Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Cited by

1. Ecosystem responses of shallow thermokarst lakes to climate-driven hydrological change: Insights from long-term monitoring of periphytic diatom community composition at Old Crow Flats (Yukon, Canada)
2. Lake and drained lake basin systems in lowland permafrost regions

View Options

View options

PDF

View PDF

Get Access

Media

Media

Other

Tables

Share Options

Share

Share the article link

Share on social media

Cookies Notification

We use cookies to improve your website experience. To learn about our use of cookies and how you can manage your cookie settings, please see our Cookie Policy.
×