Determining and Using Reference Condition in Biological Assessments II
Presiding: T Whittier, Dynamac Corporation; B Hughes, Oregon State University
NB14D-01 INVITED 15:30h
Defining the Reference Condition and Stressor Gradients: Making Black Boxes Transparent
Bioassessment of freshwater ecosystems using either the Reference Condition Approach (common in Canada, Australia, and EU) or the Multimetric Approach (common in USA) has developed rapidly, but is still limited by vague or subjective characterization of stressor gradients such as landscape degradation, added suspended sediments, and nutrient and pesticide additions. The struggle to adequately define what is really the Reference Condition in a given study, or what a credible gradient of stress is, permeates recent bioassessments. Using examples from hundreds of small watersheds in British Columbia, I will show how to catalogue all potential stressors of an aquatic ecosystem in a study area, and then characterize the multidimensional nature of the stressor gradient(s). This enables us to define the Reference Condition and stressor gradients empirically rather than arbitrarily. Perfectly analogous to characterizing multidimensional, natural gradients of environmental variability, we can relate the stressor gradients to gradients of biota and use such a relationship to predict the effects of increasing or reducing levels of stressors on the system.
NB14D-02 15:45h
An Iterative Procedure for Identifying High-Quality Reference Sites From a Set of Less-Than-Pristine Catchments.
Selection of reference sites for bioassessment purposes usually relies on determining the amount of landscape alteration that has occurred within catchments upstream of candidate sites. By definition, biota in streams draining unaltered catchments are in reference condition. However, few catchments exist that have not been altered in some way. Furthermore, such pristine catchments are usually found in unique settings, and we cannot use data from their streams to infer what reference conditions should be like elsewhere. How much can we relax the landscape criteria used to initially screen reference sites and be confident that biological integrity is still similar across sites? Once a set of high-quality reference sites is selected based on stringent landscape criteria, bioassessments will reveal what other sites in the region are equivalent to reference. These sites can be used to increase the sample size of reference sites and to some extent the range of environmental conditions represented by the reference sites. By applying this iterative procedure, we have more than doubled the number of reference sites in several regions.
NB14D-03 INVITED 16:00h
Using Taxon-Environment Relationships for Defining Macroinvertebrate Reference Communities in Disturbed Areas
The pervasive modification of river systems by humans has created a need for assessment methods that determine site-specific reference communities across such landscapes. Where the composition of taxa under undisturbed conditions (reference condition communities) can be determined for a site, the site may then be assessed by comparing the observed community with the expected community. Macroinvertebrate training data from south east Australia were used to develop regression models for the distribution of macroinvertebrate taxa associated with large-scale variables (robust to human activity) and mutable variables as candidate predictors. Natural values of the mutable variables were also predicted with process-based models such as those used for predicting sediment and nutrient loads. The regression models for taxa distributions then allow predictions to reconstruct reference condition communities by substituting process model estimates of natural values of mutable variables in place of the observed values. The predicted distribution of taxa based on estimated natural values of mutable variables can be used for biotic assessment of site condition or can be incorporated into other reference condition bioassessment approaches. The method has been tested initially in areas where reference sites are available and subsequently it is intended extended to areas with few, if any, reference sites.
NB14D-04 INVITED 16:15h
A new approach for defining reference conditions in streams
Assessments of stream fish or benthic communities normally involve a contrast of conditions at "test" sites to conditions represented by "regional"-reference sites that are either pristine or "least"-impaired. Identification of reference sites is difficult and normally involves a variety of subjective criteria. The development of reference models for stream fish and benthos in southern Ontario is particularly challenging because there are few undeveloped areas and there is no consensus on criteria for least impaired condition. Rather than identify sites as representing a least-impaired condition, we have developed a series of models that relate the existing biophysical condition of streams (i.e., the fish, benthos and instream habitat) to landscape (i.e., slope, geology, catchment area) and landcover (% imperviousness). Relationships between indices of biophysical condition and imperviousness can be used to "hindcast" or estimate the expected biophysical condition at a variety of imperviousness scenarios. The models cannot be used to predict conditions outside the calibration data range, but this approach does allow us to make use of an impairment gradient, and make predictions with a minimal number of least-impaired sites. The difference between the hindcast reference and present day conditions is an estimate of the present-day impacts.
NB14D-05 INVITED 16:30h
Hierarchical influences on the biophysical properties of natural stream channels
In this paper we focus on how catchment and proximity effects influence the biophysical properties of streams and these conditions are modified by overall landuse in the catchment and site conditions. Biophysical data (fish, inverts, instream habitat, temperature and baseflow) were collected on wadable streams flowing into the Lake Ontario basin. A GIS application was developed to characterize the landscape conditions for each site. For each of 700 -1400 sites, biophysical and GIS data (drainage area, geology, landuse, slope, stream length, and climatic conditions) were summarized. We explored the relationship between slope and drainage area and two composite indices, one the baseflow index (BFI) combined geology and drainage potential and the other, percent impervious cover (PIC) provided a single metric of disturbance on the landscape. Multivariate approaches were used to develop models to relate each metric to landscape conditions, and PIC. Residuals of this analysis were related to local conditions to determine how much of the remaining variance was explained by these conditions. Our results indicate that several metrics demonstrate a threshold response to PIC and that local conditions have minimal capacity to mitigate these effects. Below the threshold several metrics demonstrated a linear response to PIC that enables predictions to be made of effects from development in a catchment. Instream habitat conditions were more important for some biota (e.g. brown trout) than others. We demonstrate how model results can be used to classify sites or stream segments based on predicted conditions and how results can be used to generate coarse population estimates