Threatened and Endangered Habitat The Case of the

Threatened and Endangered Habitat: The Case of the Blackside Dace John Ripy Ben Blandford Jidan Duan Ted Grossardt Kentucky Transportation Center Songlin Fei Dept. of Forestry UK (Purdue)

What is the Blackside Dace?

Project Goals • Determine a better way for the Transportation Cabinet to anticipate the location of T & E Habitat for purposes of avoidance in transportation planning.

Strategy • Explore two methods of prediction – Statistical, observation-based models – Rules-based models not requiring existing data for construction

Major Components of Study • Established relationship with KSNP as repository of data and professional knowledge about T & E species and habitat. • Retained Dept of Forestry Prof. Songlin Fei as expert on statistical modeling of environmental niches. • KTC adapted existing rules-based archaeological predictive modeling approaches to habitat suitability modeling.

Tasks • Worked with KSNP to identify suitable observation data sets for Dr. Fei to use. • Used published research and KSNP expertise to devise rules-based approach to be applied to GIS. • Spent considerable effort translating rules-based themes into appropriate GIS coverages. • Provided same GIS coverages to Dr. Fei as used by rules -based approach. • Constructed models and coverages using both methods • Compared results for accuracy, coverage, etc using KSNP observation data

Executive Summary • Existing GIS data is not necessarily readily useful for habitat modeling needs • Because fairly detailed studies have been done on existing habitat for threatened species, rules-based approaches can be quite effective • Statistical models and simple rules-based models perform similarly within the spatial domain of existing observations • Unknown (yet) whether one or the other will perform better in “unsampled” areas • Rules-based models have the advantage of being potentially more useful under conditions of limited data (i. e. scarce species)

Blackside Dace Domain (known) About 70 sites

Initial Possible Habitat Factors • Tennessee Tech, Jones (2005) : – – – – Gradient Turbidity Dissolved oxygen content Water temperature Conductivity Percent riffle Link magnitude (Stream order) • Black (2007): – Conductivity – Water temperature • Kentucky State Nature Preserve: – – – Riparian vegetation type and width Presence/absence of predator fish Density of bridges and culverts over the stream network Riffle/pool ratio Density of oil and gas wells in the watershed.

Selection Considerations • GIS available to support? • Relative importance • Efficiency of model (number of factors to accommodate)

Factors Chosen for Modeling • • Gradient Canopy Riparian vegetation type Water conductivity Riparian zone width Bridges/culvert density Link Magnitude/Stream order

Gradient • Rationale: – Streams characterized by moderate to high gradient are not believed to be conducive to blackside dace habitation, as are streams of unusually low gradient. – Jones (2005), dace are most likely to inhabit streams of a gradient of one to six percent at the stream scale. – Mattingly (2005) blackside dace were four times more likely to occur in streams of a crude gradient of one to six percent. – Cumberlands quite mountainous • GIS Preparation – Data acquired from the Kentucky Geospatial Data Clearinghouse – Derived from 10 meter DEMs and mosaiced together to form a single image. – Calculated by giving a single gradient value to the entire stream segment, based on the change in elevation from the beginning of the segment to the end.

Tools for Slope • What we used – Extract Values to Points (Spatial Analyst) – Calculate Geometry – Length (or Shape. Length) • What we thought about – Slope (percent) – Focal Statistics (Range) with pre-determined Neighborhood http: //resources. arcgis. com

Five foot DEM Coverage

Extracting Data from Servers • Create Fishnet (Data Management) – Set cell size to constraints of server limits • Use a Cursor to loop through cells • Extract by mask (for each cell) http: //resources. arcgis. com

Canopy • Rationale: – Believed to directly impact the water temperature – Temperature a critical component of dace habitats (Black 2007; Black and Mattingly 2007; Jones 2005). • Black (2007): Temperature one of the most significant predictive factors for blackside dace presence/absence.

Tools for Canopy • Image Analysis – Band Arithmetic function – NDVI method • Normalized Difference Vegetation Index • Extract Bands – Calculate percentages with Map Algebra http: //resources. arcgis. com

Riparian Vegetation • Rationale: – Essential role in water quality and habitat maintenance within an ecosystem (Naiman, Decamps, and Pollock 1993). – Filter out pollutants and other sediments, minimize flood events and regulate water temperatures.

Mine Density • Rationale: – Strip mining a major cause of habitat degradation and decreasing populations of blackside dace (Biggins 1988; Black and Mattingly 2007; Mattingly 2005; Wayne C. Starnes and Lynn B. Starnes 1978). – Increased specific conductivity of the water is, in effect, a function of the dissolved solids within the water, and it provides a useful measurement for understanding water quality (Wenner, Ruhlman, and Eggert 2003). – Increased conductivity levels are known to negatively and severely impact the habitat of blackside dace (Black and Mattingly 2007).

Mine Density http: //minemaps. ky. gov

Riparian Zone Width • Rationale: – Characterized by vegetation type and width, height and bank slope (Delong and Brusven 1991). – Width indicates effectiveness of the riparian zone at filtering out pollutants and sediments and minimizing flood events. Developing Methods to Map the Region’s Riparian Areas – Ethan Inlander

Tools for Riparian Zone • Slope Degrees (less than 7) • Focal Variety (less than 5) • Path Distance (less than 225) – From streambed • Combine – Use the combined raster to clip the vegetation index to the riparian zone

Bridge / Culvert Density • Rationale – Limited mobility due to the lack of adequate connectivity of the stream network (Detar 2004) keeps Blackside Dace from reaching potentially suitable habitat. – Dams create large and deep pools with predator fish such as largemouth bass and redbreast sunfish (Mattingly 2005). – Roads over streams may yield poorly constructed culverts resulting in a small ‘waterfall’ as the water exits down the slope.

Link Magnitude / Stream Order • Rationale: – Stream order (also known as link magnitude) positively correlated with blackside dace presence/absence. • Jones (2005): stream order combined with connectivity the strongest predictive habitat model. • Blackside dace prefer small to moderate streams (Detar 2004; O'Bara 1990). – Strahler stream order captures the relative volume and accumulation of a stream segment within a network

Hydrology Tools 1 2 http: //resources. arcgis. com 3

Rules-Based Model Construction Habitat Factor 1 Gradient (stream >6% 2 4 – 6% 3 2 – 4% 4 <2% 50 – 70% 70 – 90% >90% Grass, Herbaceous, Pasture (hay) 5 – 30% Shrubs, Scrub Forested 1 -5% 0% <6 m 6 -12 m 12 -18 m >18 m Bridges/Culverts Density (per 90 sq/meters) >4 2. 5 – 4 1 – 2. 5 <1 Stream Order (Strahler) 6– 7 1 4– 5 2 -3 level) Canopy (%coverage) 0 – 50% Riparian Vegetation Cultivated, Developed, Barren Density >30% Mine (%HUC 14 area) Riparian Width Zone

Weighted Overlay GIS Layers A single stream raster was created containing data for habitat factors conducive to predicting Blackside Dace presence along stream segments • Gradient • Stream Order • Canopy • Riparian Width • Land Cover