There are several large military installations in the Sand Hills ecoregion of the Atlantic Coastal Plain that protect unique and valuable ecosystems including blackwater streams, a distinctive resource that supports a high diversity of vertebrates and invertebrates. Degradation of coastal plain streams can result in a significant loss of biodiversity and adversely affect ecological services such as nutrient cycling and natural mechanisms of water purification. The Department of Defense (DoD) is committed to the recovery of degraded ecosystems but is constrained by a lack of reference models that specify end-states representative of least disturbed conditions.

This project had three objectives: (1) develop ecological reference models that quantify key fish and benthic macroinvertebrate variables representative of least disturbed conditions in wadeable Sand Hills streams; (2) assess relationships between the reference models and important habitat variables at watershed and stream reach habitat scales; and (3) develop assessment frameworks based on the reference models that can be used to evaluate and communicate the status of coastal plain streams in relation to reference model objectives.

Technical Approach

The reference models were based on samples from stream reaches that represented least disturbed conditions, although some degraded sites were also sampled to characterize disturbance gradients. The sites were in Fort Benning, Georgia; Fort Bragg, North Carolina; Fort Gordon, South Carolina; the Savannah River Site (a Department of Energy reservation in South Carolina); state forests; a national wildlife refuge; and The Nature Conservancy lands. Digital elevation models and National Landcover Data were used to delineate watersheds, estimate disturbance levels, and calculate important watershed features. Field data were collected to characterize instream habitat including substrate, channel morphometry, snags and other structure, hydrology, and water chemistry. Fish assemblage data were collected by backpack electrofishing, and benthic macroinvertebrate assemblage data were collected using a multiple habitat sampling protocol.

Reference site screening criteria were developed to exclude disturbed sites that could bias recovery objectives. Three methods were combined to create a tiered system for definitive reference site selection that resulted in the identification of primary reference sites, which represented biological conditions of least disturbance useful for the development of reference models for full ecosystem recovery, and secondary reference sites, which represented conditions suitable for the development of best attainable models useful where prevailing or legacy land uses prevent full recovery. The reference models for fish and macroinvertebrate assemblages served as benchmarks in assessment frameworks that measured the extent to which anthropogenic disturbance caused deviations from the least disturbed state. Assessment frameworks included multimetric indices that summarized scores from selected metrics (i.e., ecologically important variables that change predictably in response to human disturbance) and predictive models that related the observed taxonomic composition to the taxonomic composition expected under least disturbed conditions. The variability, sensitivity, accuracy, responsiveness, and statistical power of both types of frameworks were measured to evaluate model performance.


Ecological reference models were developed for fish communities, macroinvertebrate communities, and stream hydrogeomorphic conditions. These models were the basis for three assessment frameworks based on fish assemblages, several assessment frameworks based on macroinvertebrate assemblages, and one assessment framework based on stream hydrogeomorphology. Habitat data were analyzed to provide background environmental information useful for interpreting biological responses and identifying factors that could affect recovery success. Data requirements for the assessment methods range from minimal to moderate and ease of implementation from simple to more complex. Performance evaluations indicated that most methods performed well at highly disturbed sites but sometimes evaluated reference sites as disturbed. They also indicated that the concomitant use of fish and macroinvertebrate methods produced more accurate assessments with greater ability to detect environmental degradation and that bioassessment accuracy was increased by using more than one assessment method for each taxonomic group since the methods differed in relative ability to detect different types of community responses to degradation.


The assessment frameworks provide an array of economical tools that can be used by DoD managers with varying resources and type of information to monitor the ecological health of wadeable streams and assess their recovery to reference conditions characteristic of the highest expectations for the ecoregion or, where necessary, best attainable conditions imposed by factors that limit full recovery. Because the frameworks represent different biotic communities as well as abiotic features, they provide a basis for comprehensive evaluations that include key stream organisms as well as the habitat conditions needed to support them. These tools can contribute to weight-of-evidence risk assessments and to monitoring programs that assess current conditions and changes associated with recovery efforts.