Using spatial imagery to understand the effects of a severe wildfire on timber rattlesnakes in the New Jersey Pine Barrens: A preliminary analysis.

Gylla MacGregor November 2007

ES771 Remote Sensing - Emporia State University

Abstract

How animals respond to environmental changes is important to organismal ecologists. Typically response to and event such as fire is measured real-time, on the ground; however, with advances in remote sensing, visual overlays using spatial imagery can be a valuable tool in demonstrating changes in habitat use. Given the range of satellite imagery analysis there is the potential to generate a rapid assessment expected animal response to disturbance. This document represents an exploratory analysis of how rattlesnakes respond to severe wildfire. Color Infrared Imagery (CIR) and on-the-ground data are used interpret the vegetation changes how such changes influenced the behavior of timber rattlesnakes.

Background and pre-disturbance studies

Timber Rattlesnake On April 4, 1995 a severe wildfire burned roughly 20,000 acres of Pineland forest in New Jersey. As a demonstration of the severity of this fire, in the hottest areas it burned through the sphagnum moss down to the water table where the snakes were hibernating. Although unharmed, some hibernating animals showed evidence of small burns on their face (H. Reinert, personal comm.). At the time of the fire, Dr. Howard Reinert (The College of New Jersey) and Robert Zappalorti (Herpetological Associates) were using radio-telemetry in a long-term study on timber rattlesnakes. Given their historical understanding of snake movements at the site prior to the fire, they expanded their monitoring to a four year study on how the fire would affect snake movement. Snakes were monitored and locations were recorded using a Trimble hand-held GPS unit for future analysis. Microhabitat variables were collected on vegetation both at snake locations and at random locations throughout the site. In terms of pre-fire behavior, the snakes had well-defined, non-overlapping movement patterns (Reinert and Zappalorti 1988). Males and non-gravid individuals used forested habitat with >50% canopy cover and a thickly vegetated understory (75%) (Reinert and Zappalorti 1988).

Analysis of aerial imagery

Image 1 is a 1991 NAPP color infrared (CIR) image of the study area (obtained from USGS, http://edcsns17cr.usgs.gov/EarthExplorer ) taken in March. Note the even coverage of deep red tones. In CIR imagery, red tones indicate vegetated areas; this image shows a well vegetated pine-oak forest. White sand roads are clearly identifiable and a branch of the stream can be seen (black line) running through the core study area. Along portions of the stream are deeper red tones reflective of the cedar swamp. Less obvious white lines are visible in the northern portion of the image. These are fire breaks which are common throughout the Pine Barrens. The fact that they are difficult to see in this 1991 image is another indication of the thick vegetation at the site. Other landmarks include a lake, appearing black in color, visible in the south east corner of the map. Image 2 was taken in the unburned section of the study site and is representative of the area prior to the 1995 wildfire.

Image 1. 1991 NAPP CIR image of the study site. The actual study area is boardered by the three roads forming a triangle.

Image 2. This photo represents what the study area looked like in the years preceeding the fire. Note the thick vegetation as reflected in the aerial image.

Image 3 is a 1995 DOQQ CIR image of the study area (obtained from New Jersey Image Warehouse, http://njgin.nj.gov/OIT_IW/index.jsp ). This image was taken 3 days following the fire. The extent of the fire is evident in this image by the large blue area. Unburned sites reflect the standard red-tones of vegetation. Notice the pockets of apparently untouched vegetation within the main study area and the nearly black regions which reflect more severly burned locations. The mix of red and blue in the north east section suggests that while this region was burned, burning was not as severe, probably due in large part to concentrated efforts to at least keep the fire away from the main road and houses. Sand roads are more apparent in this image than in the 1991 image and because the fire thinned the vegetation, the fire brakes are very obvious as compared to Image 1. Given the level of destruction at the time the image was taken, snakes might be expected to significantly alter their behavior due to significant changes in habitat. Image 4 is a photo of the site 1 year following the fire.

Image 3. 1995 NAPP CIR image of the study site. Note the devestation (dark blue) in the core study area.

Image 4. This photo represents what the study area looked like in 1996, one year following the fire.

Image 5 is a 2002 CIR digital orthophotograph of the site and was obtained from New Jersey Image Warehouse (see link above). The distribution of red across the site is supportive of ground data suggesting the variable nature of vegetation regrowth, and the gradual return of overstory and understory vegetation. Surface data showed a general inability to easily distinguish between burned and unburned sites. In fact, in some areas the only sign of a fire was the persistent black soot on the pine trees. Notice that the fire breaks are barely visible; similar to the 1991 image. This represents the lack of fire intensity and accordingly, the more rapid return to pre-fire conditions. In contrast, vegetation in the heart of the study area still remains thin in spots, with some areas demonstrating a more rapid return to pre-fire conditions than others, particularly along the stream corridor and in low-lying areas such as the southwest corner. A continued reminder of the most severely burned locations can be seen in the small patches of blue-gray as is observed along portions of the stream in the core study area. With the habitat approaching a return to pre-fire vegetation, the site would appear to be much more suitable for snakes. Notice though, in Image 6 how the cedar swamp still appears devistated, several years following the fire. This corresponds to some of the blue-gray regions in the image. Clearly under extreme conditions the cedar trees were permanently destroyed and thus represented a significant change in habitat.

Image 5. 2002 CIR image of study area. Although evidence of the fire clearly remains, a gradual regeneration is apparent in this image.

Image 6. Taken in 2006 this image of the severly burned cedar swamp still shows strong evidence of the fire.

Analysis of post-fire ground data

A vegetation analysis conducted one year following the fire suggested that regrowth was variable among habitats (MacGregor 1998) this was largely a function of variability in fire severity. Although canopy cover was relatively unchanging following the fire (42.8%) the lost understory (47% vegetated understory in 1995) rapidly returned to pre-fire levels (79% in 1998) (Reinert and Zappalorti 1998). These numbers are reflected in the imagery above and the lack of observable overstory in 1995 may be easily explained by the date of the fire and the date the image was collected. Many of the pitch pines appeared completely destroyed following the burn but rapidly regenerated in the spring and summer. Overall the snakes were found to be largely unaffected by the fire although there was a small change in microhabitat use immediately following the burn (Reinert and Zappalorti 1998, 1999), in fact, rattlesnakes were so dedicated to their movement patterns that they were frequently found at the same location in successive years (MacGregor 1998). The only obvious impact was in the severely burned areas of the dens. Snakes utilizing these sites prior to the fire returned to the burnt den each fall of the study but did not stay; rather they denned in unburned locations (Reinert and Zappalorti 1998).

Blending ground data with satellite imagery – future possibilities

The overlay of snake movements (see Map 1 below) provide a clear visual as to the largely unchanging movement patterns. Microhabitat changes immediately following the fire are not obvious in the current aerial overlay; however, the use of Landsat satellite imagery and vegetation analysis would likely reflect more the detailed vegetation changes that drove habitat selections. The most obvious and critical aspect of the fire was the destruction of an active den. This represents a significant alteration in habitat that although not apparent in this preliminary analysis, would likely become evident in a more detailed in Landsat or SPOT2 image analysis through the use of the Normalized Difference Vegetation Index (NDVI), Differenced Normalized Burn Ratio (dNBR), Composite Burn Indices (CBI) or similar indices (Cocke et al 2005; Hammill and Bradstock 2006; Key and Benson undated). These indices are valuable in that they can assign a numerical value to vegetation changes. This value may then be statically compared to values both before and after the fire allowing a more detailed measure of change.

Using this on-the-ground understanding of an animal’s response to fire intensity, and comparing this to satellite image analysis can provide a significant baseline data for future analysis. Studies may reveal that there is a statistical value range that represents the degree of potential negative impact. Similar values may then be calculated for any burned location for a rapid assessment of wildlife impact. This may be particularly relevant in areas where there is active management of threatened or endangered species or in areas where fire management plans are in place or pending.

Overlay of two timber rattlesnake movement patterns with 1995 (left) and 2001 (right) spatial imagery.

Map 1. This map compares 1995 and 2002 imagery with representative individuals. Because there is known fidelity to movement patters, snake movements not captured the year of the fire would still reflect the individuals typical movement patten. This is also demonstrated in this map using three year's worth of movement patterns. The 1995 image on the right shows the 1996 and 1997 movements for Snake A in yellow (1996) and green (1995). White triangles represent 1997 movements of Snake B. The 2001 imagery shows the 1998 movements for both Snake A (yellow) and B (white). Notice how the individuals continue their utilization of burned areas.

References

Cocke, A. E., P. Z. Fule and J. E. Crouse. 2005. Comparison of burn severity assessments using differenced normalized burn ratio and ground data. International Journal of Wildland Fire. 14(2):189-198.

Hammill, K. A. and R. A. Bradstock. 2006. Remote sensing of fire severity in the Blue Mountains: influence of vegetation type and inferring fire intensity. Internal Journal of Wildland Fire 15(2):213-226.

Key, C. and N. C. Benson. The normalized burn ratio (NBR): a Landsat TM radiometric measure of burn severity. http://www.nrmsc.usgs.gov/research/ndbr.htm

MacGregor, G. A. 1998. The impact of fire on selected small mammals species in the Pine Barrens of New Jersey. The College of New Jersey Journal of Student Scholarship. 1:33-42.

MacGregor, G. A. 1999. Foraging behavior of the timber rattlesnake (Crotalus horridus) in the New Jersey Pine Barrens. Masters Thesis. Clarion University of Pennsylvania.

Reinert, H. K. and R. T. Zappalorti. 1988. Timber rattlesnakes (Crotalus horridus) of the Pine Barrens: their movement patterns and habitat preference. Copeia 1988:964-978.

Reinert, H. K. and R. T. Zappalorti. 1998. The impact of fire-altered habitat on the ecology of timber rattlesnakes (Crotalus horridus) in the Upper Cedar Creek watershed of Ocean County, New Jersey. Research Progress Report to the Barnegat Bay Environmental Group.

Reinert, H. K. and R. T. Zappalorti. 1999. The impact of fire-altered habitat on the ecology of timber rattlesnakes (Crotalus horridus) in the Upper Cedar Creek watershed of Barnegate Bay, Ocean County, New Jersey. Research Progress Report to the Barnegat Bay Environmental Group.