Abstract
Human access to surface water resources in the Northern Great Plains (NGP) is challenged by availability and quality, and ecosystem health objectives for these characteristics have not been well developed. Here, we present a predictive multivariate model using the reference condition approach to inform goals for ecosystem health assessment. Benthic communities and abiotic variables were collected at 280 potential reference sites and 8 test sites, and of these, reference sites with least amount of human activity (n = 83) were classified into three community groups and summary metrics. Discriminant function analysis and cross-validation determined that stream order and ecoregion predicted 68.7% of the sites correctly, thus enabling comparison of sites with unknown condition to reference site groups. We then evaluated metrics through Test Site Analysis and stressor gradient analysis in each biological group. Beetle and amphipod fauna were found to be important for condition assessment in addition to traditional metrics of species richness, abundance, detritivory, Ephemeroptera/Plecoptera/Trichoptera dominance, and assemblage composition. These results provide least disturbed reference condition and ecological insights into land use impacts in the NGP. Ultimately, this model is an effective tool for evaluating biotic condition, enables prioritizing river management strategies, and can quantify the efficacy of mitigation measures.
Introduction
Water security is threatened globally by stressors brought about by resource extraction, waste water, widespread land cover change, urbanization, and engineered works such as reservoirs (Vörösmarty et al. 2010). The development of methods to monitor the impact of these anthropogenic perturbations efficiently and effectively are crucial to conserve ecosystem health for future use (Vörösmarty et al. 2010). Communities of aquatic organisms and the traits they possess are often used to evaluate the biological condition of freshwater habitats as they are responsive to changes in environmental characteristics and express the ultimate ecological consequences of anthropogenic perturbation (Resh and Jackson 1993). Benthic macroinvertebrates in particular have been used in biomonitoring studies since the early 1900s when summary metrics of their sensitivity to organic pollution were incorporated into a European-focused Saprobien system of river condition (Kolkowitz and Marsson 1908). Since this initial application, many multimetric (e.g., Karr et al. 1986; McCormick et al. 2001; Klemm et al. 2003), Nearest-Neighbour Analysis (e.g., Linke et al. 2005), Bayesian (e.g., Qian et al. 2003), and multivariate (e.g., Wright 1984; Hawkins et al. 2000; Reynoldson et al. 2001; Bowman and Somers 2006) approaches to biomonitoring have been developed and applied globally with the intention of quantifying ecosystem health in measures that can be communicated among environmental managers and the public. However, regardless of method used, the definition of biological condition and the ability to assess impairment from healthy condition relies on an understanding of what community of organisms and their traits are expected in the absence of human activity (Ode et al. 2005). These baselines are categorized commonly as reference condition in biological monitoring, and the method of evaluating condition against reference condition has been labeled as the reference condition approach (RCA; Wright 1995).
A fundamental challenge in the RCA, or any control baseline in environmental management for that matter, is to define a reference benchmark such as historical condition, least disturbed condition, minimally disturbed condition, or best attainable condition (Stoddard et al. 2006). Historical condition is a reference prior to human activity, least disturbed accepts some human activity and seeks the best available condition, while minimally disturbed condition requires only minor levels of human activity, and best attainable condition is the most optimistic condition given mitigation intervention at least disturbed sites (Stoddard et al. 2006). Although historical, minimal and best attainable conditions can be appealing for their relevance to what condition could be in the absence of human activity, on a practical level, many regions such as the Northern Great Plains (NGPs) do not yet have adequate historical condition assessment, representation of minimally impacted sites, nor examples of best attainable conditions given mitigation solutions (Stoddard et al. 2006). Therefore, the least disturbed condition is likely the most realistic characterization of reference ecosystem health possible today.
Aquatic macroinvertebrate community (species composition, abundance of individuals, and species diversity) change depends on natural characteristics of hydrology and fluctuation of flow condition (Poff et al. 1997), as well as the type and amount of water chemistry and pollution present in the system (Hilsenhoff 1988). Measurable, predictable change enables development of ecological health metrics. Other jurisdictions (e.g., United States, Britain, European Union, and Australia) have developed macroinvertebrate aquatic health measures that together with water chemistry, are used to establish and monitor surface water quality. However, macroinvertebrate measures for the NGPs have been more difficult to develop, because the natural extremes (e.g., winter, drought, erosion, high nutrient and productivity; Covich et al. 1997) of this region result in a community characterized by taxa already more tolerant to abiotic extremes than those found elsewhere in North America (Williams 1985). Therefore, macroinvertebrates in the NGP may be insensitive to pollution relative to other regions. For example, benthic metrics such as % EPT developed to respond negatively to increasing turbidity elsewhere in North America (Roy et al. 2003; Freeman and Schorr 2004; Evans-White et al. 2009) have been found to respond positively to increasing turbidity in the NGPs due to differing composition of taxa in the metric (e.g., burrowing mayflies, Phillips et al. 2016). As such, feasibility of an aquatic health measure using NGP macroinvertebrates requires investigation before existing metrics from other jurisdictions are applied, and these metrics must be validated for particular stressors to account for variability across ecoregions.
Rivers serve as the principal source of renewable freshwater supply for humans and freshwater ecosystems globally (World Water Assessment Programme 2009) and are of great importance in the NGPs (Dodds et al. 2004). Despite this, the NGPs in western Canada possess a landscape largely devoid of unaltered historical conditions and the tools to assess impacts of potential anthropogenic perturbations of lotic aquatic ecosystems have not been developed. Intensive agriculture, water management, and development have impacted most rivers and streams since European colonization, and the stressors these activities impart do not have ecosystem health measurement tools to reflect the impact they may be having. Here, we developed a benthic macroinvertebrate-based, multivariate biomonitoring tool using least disturbed reference conditions and Test Site Analysis (TSA) condition assessment for wadeable rivers and streams in the NGPs. The steps in its development are as follows:
1. Characterize scope of stream ecosystems and human stressors to include in the study.
2. Define least disturbed condition based on stressor gradients.
3. Classify biological communities at reference sites and determine what underlying abiotic characteristics best discriminate between classifications. These relationships are used to match test sites of interest to appropriate reference groups.
4. Select summary metrics of the benthic macroinvertebrate community that capture significant community structure and eliminate redundant descriptors.
5. Designate test sites to appropriate reference groups using predictive modeling and evaluate biological condition based on metrics identified above using TSA.
Our goal was to apply a RCA based on benthic macroinvertebrate communities to develop an ecosystem health monitoring tool for NGPs rivers that can be applied to assess the effects of anthropogenic perturbation. This study develops a predictive model to evaluate the amount of variation in the benthic macroinvertebrate communities that can be explained by abiotic characteristics of rivers and streams at reference sites with the least amount of human activity. Further, we apply these abiotic characteristics to assign test sites to appropriate reference benchmarks and evaluate biological condition based on community metrics to predict what ecosystem health metrics should resemble under the least amount of human activity.
