Landscape measurements

Get Complete Project Material File(s) Now! »

Methods Study Site Selection

A large portion of the Delmarva fox squirrel’s range is inaccessible to public access due to private property status or lack of access routes. Chesapeake Marshlands National Wildlife Refuge not only offered high accessibility to a large area of potential DFS habitat, but Delmarva fox squirrels also appear to be most abundant in this southern portion of their known range (Ratnaswamy et al. 2001). Thus, the relatively high probability of failing to detect individuals during surveys for a naturally rare species is decreased. Therefore I assume that the lack of use of certain habitat types by Delmarva fox squirrels in this area was purposeful, rather than resulting from low density distribution.
Like other fox squirrel species, Delmarva fox squirrels are known to use non-traditional cover types such as crop fields and mast-producing pine thickets as supplemental feeding areas (Weigl et al. 1989, Kantola 1986, Nixon et al. 1980, Taylor 1976). Should records of such use be included in a process intended to generate a predictive model used to identify DFS habitat, the current definition of habitat would be broadened considerably. Under federal law (Endangered Species Act, 87 Stat. 884, as amended; 16 U.S.C. 1531 et seq.), areas considered DFS habitat must be protected. To maintain reasonable and prudent regulatory applicability of my results, I limited biological inference to cover types that have been shown capable of supporting a breeding population of Delmarva fox squirrels.
I used a random sampling approach to select forested study sites within CMNWR. The refuge had established a 500 meter spaced grid in 2003 to sample baseline flora and fauna conditions refuge-wide (Figure 3). During the 2003 effort each gridpoint was located with GPS, and a metal post with a site-specific alpha-numeric identifier was placed at each location. Using GIS maps (Arcview GIS 3.3) overlaid with the GPS grid, I selected all gridpoints previously identified as forest by MD DNR GAP analysis (Rasberry et al. 2003). Of the 400 total gridpoints on the refuge, 134 were determined by DNR GAP to be within forested habitat (Table 2). Of these 134 points, 126 were visited and 86 were sampled. The 48 forested points that were eliminated from this study did not qualify because they were either misidentified (n = 6), inaccessible (n = 8), regenerating clearcuts (n = 5), or no longer viable forests due to saltwater inundation (n = 29).
I collected preliminary vegetation data on 25 sites from June 2004 to August 2004. I collected complete vegetation data and DFS use data on all 86 sites from September 2004 to December 2004.

Vegetation Sampling

Field measurements followed Noon’s (1981) sampling protocol for forest habitats (Table 3). This comprehensive approach provided a complete representation of current site conditions. This also ensured I did not limit myself to a′ priori supposition, thus allowing more freedom and opportunity during my post hoc analysis.
I measured species composition, overstory, understory, shrub cover, and ground cover within a 0.04 ha circular plot (11.3 m radius) centered at each sampling point (Figure 4). Although some preliminary habitat data were collected between June 2004 and August 2004, all measurements that were seasonally affected (i.e., shrub and groundcover variables) were recorded concurrently with DFS sampling. Though no data were available to support the supposition that DFS may stop or start using an area due to seasonal fluctuation in vegetation structure, this coordination of measurements ensured we did not risk this sampling incongruity. I summarized field data to generate variables used to build habitat models (Table 4). Habitat data sheets are presented in Appendix I.
Shrub stem types were separated into 7 categories (Figure 5). Sparse grasses (< 100 blades/ m2) comprised stem type A, single bole shrubs over 15 cm (eg: high bush blueberry, pepperbush) comprised stem type B, single bole shrubs < 15 cm (eg: low bush blueberry) comprised stem type C, briar (eg: greenbrier, raspberry) comprised stem type D, short turf-like grasses comprised stem type E, and tall marsh grasses comprised stem type F. Stem type G was formed by pepperbush with regenerating seedlings evenly distributed across the ground. Stem types A – C have an irregular growth pattern, stem type D is generally clumped or irregular, and stem types E – G are usually evenly distributed.
Measurements of water depth and dispersion, log cover, snag size, and snag density were recorded on 80 of the 86 sites. However, tree mortality resulting from approximately a decade of saltwater intrusion confounded these variables. As a result, these variables were eliminated from the final analysis.

Landscape measurements

The position of a potential habitat parcel within the landscape could very well influence whether or not it is used by Delmarva fox squirrels. Therefore, I included landscape measurements such as distance to nearest agricultural field, distance to nearest road, patch size, and stand size in my analyses (Table 3). I defined a forest patch as a contiguous group of trees that was sufficiently uniform in species composition, arrangement of age classes, and condition to be a homogeneous and distinguishable unit (Smith 1962). A forest stand was defined as a contiguous group of patches with no apparent barriers to DFS movement. Distance to a road or agricultural field was measured pragmatically (i.e., not across barriers to DFS mobility such as wide water bodies or large expanses of marsh).
I used the Delmarva fox squirrel GIS database currently maintained by the USFWS Chesapeake Bay Field office to determine landscape measurements. Measurements were taken using digital ortho quarter quads (DOQQs) and the ArcView geodesic distance and area measurement tool. A precision of 100 linear feet and 5 acres is assumed.

Photomonitor Procedure

Delmarva fox squirrel presence was determined using an infrared photomonitor at each survey point. Photomonitoring provided a means of detecting Delmarva fox squirrels without the time, cost, labor or stress to the animal involved with live trapping. Though population estimates cannot be obtained from photomonitoring survey efforts because of the similarity in pelage among individuals of this species, basic presence data can be confidently secured using this technique.
A photomonitor was placed at the center of each 0.04 ha study plot for 7 days. The area in which vegetation sampling occurred (0.04 ha) was approximately 0.3 % of the estimated Delmarva fox squirrel home range (12.78 ha; Paglione 1996). Based on the 500 m spaced grid, the area associated with each gridpoint was 25 ha. I assumed the photomonitor’s area of sensitivity did not cover this entire 25 ha, but rather was limited to areas from which the bait could be sighted by a squirrel. Because Dorchester County has almost consistently flat topography, the photomonitor’s area of sensitivity thus only varied with understory density.
Trailmaster® brand (Goodson & Associates, Inc., Lenexa, Kansas, USA) TM1500 and TM1550 active monitors were used in this study. These monitors detect and photograph anything that breaks an invisible infrared beam (beam length < 45 meters) emitted from a transmitter to a receiver (Figure 6). Units were usually secured to trees within 11.3 m of the central marker post. If no trees of the proper diameter (10 – 30 cm dbh) or spacing (1.5 – 5 m apart) were within the study plot , 5 cm x 10 cm x 1m stakes were used to secure the receiving unit. Beam length at the sampling points was < 5 meters to decrease the opportunity for a non-target animal to break the beam. The beam was set 7-9 cm from the ground to ensure contact with the body of the squirrel. A weather resistant 35 mm camera (TM-351) with 400 ISO film was used for the photographs. For this project, the monitors were programmed to register all events but only photograph those occurring between 0500 and 2000 hours. Though Delmarva fox squirrels are known to be almost strictly diurnal (C. Morris, USFWS, unpublished data), I attempted to capture any crepuscular activity as well. Both the monitors and the cameras recorded events by date and time to the minute. The receiver was programmed to register an event once an animal interrupted the beam for at least 1/4 second. This restriction reduced photographs of falling vegetation or fast-moving deer. The TM-351 camera had a 5 minute delay between triggered photographs to reduce the possibility of an individual animal visit exhausting an entire roll of film.
A standard squirrel sized (48.26 cm x 15.24 cm x 15.24 cm) Tomahawk live trap (# 103, Tomahawk Live Trap Co., Tomahawk, WI) was wired open and placed at a right angle to the transmitter to direct squirrel movement in front of the beam. The trap also acted to reduce bait theft by raccoon and deer. The bait (cob corn) was partially obstructed with dowel rods or sticks to inhibit a non-target species (e.g., raccoon, deer, etc.) from immediately removing the bait. This impediment also indirectly produced more photographic opportunities as a squirrel circled the trap to investigate alternate means of reaching the bait.
Photomonitoring units were placed at the sampling point for 7 days between September and December, 2004. Units were only on each site for 7 days to ensure I was obtaining records of normal use rather than attracting DFS to a site with bait. Vegetation characteristics were measured at each point upon collection of the photomonitoring units; this allowed accurate representation of the sampling site at the time DFS use was ascertained, while assuring the human activity associated with sampling vegetation had no effect on DFS response. No photomonitoring data were collected during summer months because of a period of perceived inactivity in DFS, supposedly due to higher temperatures (Paglione 1996). Photomonitoring data sheets are presented in Appendix II.

Reasons for Decline
Habitat models
Recovery Goals
Study Site Selection
Vegetative Sampling.
Landscape measurements
Photomonitor Procedure .
A Priori Model Development.
Trap Results
Habitat characteristics
Shrub cover
Ground cover
Model Development
Integrated Post Hoc Model.
Currently used DFS Model
Endangered Species Overview
Delmarva fox squirrel habitat models
Modeling Overview
Current Hypotheses Supported by Results of this Study
Building a Predictive Model of Delmarva Fox Squirrel (Sciurus niger cinereus) Occurrence Using Infrared Photomonitors

Related Posts