USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Population and Movement Characteristics of Radio-Collared Striped Skunks in North Dakota During an Epizootic of Rabies

Materials and Methods

The study was conducted 50 km north of Jamestown, Stutsman County, North Dakota in the Southern Drift Plain of the Prairie Pothole Region (98° 52'N, 47° 12'W). Landscape of this area is gently rolling and was nearly two-thirds cropland; remaining land was mostly grassland and wetlands. The study area was about 93 km2 and contained 15 occupied farmsteads. The Arrowwood National Wildlife Refuge managed by the U.S. Fish and Wildlife Service is located along the entire east side of the study area, intersected by the James River. Drought conditions prevailed during 1991 and 1992, and except for the James River and a few impoundments and spring-fed dugouts for livestock, most wetlands in the study area were dry.

We captured striped skunks in livetraps (24 × 24 × 66 cm) (Tomahawk Live Trap Company, Tomahawk, Wisconsin, USA) and anesthetized them with 0.75 to 1.0 ml of a premixed combination of 60 mg ketamine HCL (Fort Dodge Laboratories, Fort Dodge, Iowa, USA) and 15 mg xylazine HCL (Mobay Corporation, Animal Health Division, Shawnee, Kansas, USA) per ml (Crabtree and Wolfe, 1988). We placed animals in shade near the capture site for recovery. If euthanasia was necessary, we administered a combination of pentobarbital sodium (390 mg/ml) and phenytoin sodium (50 mg/ml) (Shering-Plough Animal Health Corporation, Kenilworth, New Jersey, USA) after anesthesia. These and all other protocols were approved by the Northern Prairie Science Center Animal Care and Use Committee and followed recommendations of The American Society of Mammalogists (1987). Field personnel who handled skunks had received prophylactic rabies immunization.

In 1991, we captured skunks in the interior 44 km2 of the study area so we could test experimental protocols of the food provisioning study. Capture was primarily on two occasions (18 to 20 April and 21 to 31 May). We recaptured and euthanized animals approximately 3 to 4 wk after capture so we could remove those that had been tested from the population and so we could determine animal ages.

In 1992, we captured skunks throughout the 93-km2 study area, primarily during 31 March to 5 April. We trapped and equipped new skunks with radio transmitters on several occasions to replace animals that died and maintain sufficient marked animals for population monitoring. In mid-July, to permit age determination and testing for rabies, we euthanized and retrieved all radio-collared skunks and some unmarked skunks captured at sites where we provided supplemental food (hereafter called feeding sites). During both years, all nontarget animals were released unharmed.

At initial capture we determined the sex of each skunk and placed a numbered tag (1 × 3 mm) in each ear. At the same time we attached an approximately 60 g radio-collar (164-167 Mhz) (Advanced Telemetry Systems, Inc., Isanti, Minnesota, USA) detectable up to 3.2 km; collars contained a motion-sensor that permitted us to identify mortality (animal motionless for ≥3 hr). We noted and recorded obvious scars and bite marks on skunks each time an animal was anesthetized. We examined live females for pregnancy by palpation and conducted post mortem examinations for fetal swellings or fresh uterine implantation sites; lactation also was interpreted as evidence of pregnancy in the current year. We estimated skunk ages post mortem to the nearest year by counting annuli in canine teeth (Johnston and Watt, 1981).

In 1992, we provided foods (chicken eggs, dry dog food, fish, sunflower seeds) and an attractant (fatty acid scent disk, U.S. Fish and Wildlife Service, Pocatello Supply Depot, Pocatello, Idaho, USA) at 10 permanent feeding sites established ≥3.2 km apart (Fig. 1). Each site was composed of four 1-m-diameter plots (20 m apart in a row) where foods and lure were provided continuously from 1 April to 9 July. Foods were replenished daily between 0800 and 1500 and the attractant was replaced every 3 to 4 wk. Sites were in idle grassland ≥50 m from the nearest road to avoid disturbance by humans.

Observers determined locations of radio-collared skunks by triangulation from vehicles equipped with null-peak antenna systems. Locations were plotted immediately on a 1:24,000 map and converted later to universal transmercator (UTM) coordinates. Locations were not recorded during the first 10 to 12 hours after each occasion when a skunk was anesthetized. Accuracy of locations was confirmed by sightings of radio-equipped animals (Fenn and Macdonald, 1995) and during visits to retreats occupied by skunks and locations of carcasses; accuracy was usually within 50 m.

We located skunks once each day between 0700 and 1800 to determine their diurnal retreats (hereafter called day-tracking). We located skunks at 2-hour intervals between 1800 and 0700 the next morning to determine their nocturnal movements (hereafter called night-tracking). During 18 April to 20 June 1991, we conducted day-and night-tracking irregularly while we tested study protocols. From 15 April to 9 July 1992, we systematically conducted day-tracking on 6 of every 7 days and night-tracking on 5 nights every half month. Nights were preselected at the beginning of each half-month period to distribute tracking effort throughout the period. During night-tracking, two observers each monitored one half of the radio-collared skunks. Any radio-collared skunk that was not located during day-or night-tracking was sought continuously during all tracking efforts, including occasional searches outside the study area. Animals that dispersed from the study area were recaptured when found, and euthanized.

Skunks suspected of being sick or dead, based on radio-signal or unusual location, were checked promptly the next morning by an observer on foot with a hand-held radio-receiver. Skunks found alive but immobile were euthanized.

In 1992, euthanized skunks and those found dead were transported to Jamestown for necropsy. Upper canine teeth were removed and frozen until used to estimate animal age. Brains were removed, placed in individual identified containers, and shipped under refrigeration to the North Dakota State University Veterinary Diagnostic Laboratory.

One half of each brain was fixed in 10% buffered formalin, stained with hematoxylin and eosin (H&E), and examined by light microscopy. Impression smears were made from the unfixed half of the brain, stained with fluorescein labelled antibody (FA), and examined by fluorescent microscopy (Velleca and Forrester, 1981). A positive diagnosis for rabies was based on a positive FA test, with or without the presence of lesions on the H&E-stained slides.

Brains with lesions of viral or lymphocytic encephalitis, but negative for rabies antigen by the FA test, were injected intracranially into anesthetized mice. Brain tissue from treated mice was subjected to FA testing. Skunks positive for rabies by mouse inoculation test were included with others diagnosed as positive. Hereafter, we refer to skunks that tested positive for rabies antigen as rabid skunks and to those in which we did not detect rabies antigen as healthy skunks.

For animals in which we did not detect rabies antigen in mouse brain tissue by the FA test, the original skunk brain was tested again in an attempt to determine etiology of the lesions. The fixed sections were immunohistochemically stained for morbillivirus antigens using a rabbit polyclonal antisera to human measles virus nucleoprotein and an aviden-biotin complex technique (Haines and Clark 1991).

We grouped skunks into three classes by age: age class 1, <1 yr old; age class 2, ≥2 yr old and <3 yr old; age class 3, ≥3 yr old. Grouping reduced the effect of errors in age estimation, which tend to increase with animal age (Johnston et al., 1987) and also reduced statistical problems associated with small samples. We contrasted, by sex and age class, proportions of skunks captured during 1991 and 1992 with a larger sample of striped skunks obtained during the months of April to July, 1979 through 1990 in similar habitats of east central North Dakota and west central Minnesota (USA) (Greenwood and Sargeant, 1994). We tested, with chi-square statistics, the effects of year (1979 through 1990, 1991, 1992), sex, and year by sex interaction on age class of skunks, using weighted least-squares estimation techniques (Grizzle et al., 1969) with the procedure CATMOD (SAS Institute, Inc., 1989a). Pairwise comparisons were conducted for significant effects using contrast statements in the procedure CATMOD. Throughout this report, variation is expressed as standard error (Mean of X± SE); differences are considered significant at P = 0.05.

We estimated survival rates (Survival Rates) of radio-collared skunks for 1991 and 1992 with Kaplan-Meier estimation techniques in a staggered entry design (Pollock et al., 1989). We computed mortality rates due to three specific causes: rabies, suspected predation by carnivores (includes some rabid skunks), and other; other includes vehicle collisions, shooting, and unknown causes. We censored (Lee, 1992) observations of healthy skunks on the date they were euthanized. For this analysis, to compensate for small sample sizes in some categories, we combined age class 3 with age class 2. We then compared Survival Rate between sexes and age classes (1 and 2) using the procedure IML (SAS Institute, Inc., 1989a) following the methods described by Sauer and Williams (1989); we considered the day of capture to be the first day an animal was at risk. We assumed that the variance of Survival Rate, when Survival Rate= 1.00, was equal to a pooled group estimate.

Because skunks may exhibit heightened levels of activity when rabid (Charlton et al., 1991 ), we examined characteristics of movement and use of space to see if we could detect differences between healthy and rabid radio-collared skunks in 1992. For individual animals, we estimated minimum rate of travel in a night, minimum distance traveled in a night, and minimum size of home range for approximate half-month periods from mid-April to mid-July: period 2, 19 April to 2 May; period 3, 3 to 15 May; period 4, 16 to 29 May; period 5, 30 May to 12 June; period 6, 13 to 25 June; period 7, 26 June to 11 July. We did not derive estimates for animals in 1991, or for period 1 (1 to 18 April) in 1992, because radio-tracking records were incomplete. We used least-squares means and their standard errors (Milliken and Johnson, 1984) to estimate population means; least-squares means are unbiased estimates of population means.

Rate of travel (m/hr) for a skunk was estimated as the distance (m) between two consecutive locations during night-tracking divided by the intervening time interval (hr). We estimated the minimum rate of travel for each radio-collared skunk for which we obtained two or more consecutive locations separated by >1 but <3 hr after it left its daytime retreat. On each night for each skunk that met these criteria, we pooled rates for individual pairs of locations and divided the pooled rate by the number of pairs of locations that night for that skunk. We pooled the estimated mean rate of travel per night for each half-month period and divided the pooled mean by the number of nights tracked within each half-month period for each animal.

Distance traveled (m/night) was estimated as the sum of the distances between consecutive locations during night-tracking. We estimated the minimum distance traveled in a night for each skunk that was tracked through a complete night (located in every 2-hr interval) after it departed its daytime retreat until it reached its daytime retreat the following morning. For each animal that met this criterion, we pooled the estimated distances traveled each night for each half-month period and divided the pooled distance by the number of nights to estimate the mean minimum distance traveled per night during a half-month period.

We estimated home range size (km2) by the minimum area method (Mohr, 1947), using SAS code reported in White and Garrott (1990). We estimated home range size for each animal for each half-month period, based on all capture and radio-tracking locations during that period. We pooled the estimates of home range size for each half-month period and divided the pooled estimate by the number of periods to estimate the mean half-month home range size.

We assessed effects of sex, half-month period, and their interactions on estimated rates of travel, distances traveled, and home range sizes of healthy and rabid skunks with a repeated measures analysis of variance (ANOVA) technique and the general linear models procedure (GLM) (SAS Institute, Inc. 1989a). Animal within sex was the whole unit, and the same animal at different half-month periods (periods 2 through 6) was the subunit (Milliken and Johnson, 1984). We compared estimates only for half-month periods 2 through 6, because we had no data for period 7 for rabid males. For the home range ANOVA, we weighted estimates for each half-month period by the square root of number of locations for that period, to compensate for differences in sample sizes. Home range size tends to increase with number of locations (White and Garrott, 1990), and there was a slight correlation between home range size and number of locations (r = 0.3, P < 0.01). We used Fisher's protected least significant difference (LSD) (Milliken and Johnson, 1984) to make pairwise comparisons among means for significant main effects.

We were interested in determining if there was more variation in rates of travel among individual skunks, among nights, or within nights throughout the summer. We evaluated sources of variation in observed rates of travel for each combination of sex and rabies status of an animal (rabid or healthy), using a variance component approach (Box et al., 1978). The design was a nested structure (Box et al., 1978) in which rates of travel were estimated at multiple times within each night for individual skunks, across multiple nights for individual skunks, and among individual skunks. Our primary interest was to examine the percent of variation explained by each component relative to the total variation for each combination. We were not interested in testing for differences in variance components among combinations of sex and rabies status. We used the MIVQUE0 method of procedure PROC VARCOMP of SAS (SAS Institute, Inc., 1989a) to compute variance components.

We attempted to isolate effects of rabies on movements of skunks during the clinical period, which we defined as the 14 days preceding death or euthanasia of an immobile animal. Charlton et al. (1984) observed a mean (± SD) duration of the clinical period of 9.7 ± 3.0 days in rabid captive skunks, but in wild skunks this stage is less clearly defined and ranged from 1 to 18 days (Charlton et al. 1991).

We computed the minimum rate of travel, minimum distance traveled, and home range for rabid skunks for the 14-day interval preceding death (hereafter called clinical period) and the previous 14 days (hereafter called pre-clinical period). We pooled results by 14-day intervals and computed means for those intervals.

We used an ANOVA to assess effects of sex, period (pre-clinical or clinical), and their interaction on estimated parameters of travel and home range size. We used the GLM procedure in a repeated measures design (Milliken and Johnson, 1984). Animal within sex was the whole unit and the same animal at different periods (pre-clinical or clinical) was the subunit. If we detected no significant interaction among main effects, we used Fisher's protected LSD (Milliken and Johnson, 1984) to make pairwise comparisons among means for significant main effects. This method of computation permitted us to standardize effects in relation to the clinical period of rabies, which was our primary objective; however, it reduced our ability to examine seasonal effects of rabies. We limited this analysis to animals observed continuously for more than 28 days.

We evaluated the locations of deaths among rabid skunks in 1992 to determine if deaths were random or clumped among the radio-collared population. We included in this analysis all rabid skunks found dead, skunks that we discovered immobile in the late clinical stage of rabies, and skunks with rabies that we believed were killed by predators. We calculated a single location (hereafter called activity center) that represented the home range of each radio-collared skunk tested for rabies. The activity center was based on the mean UTM coordinates of all capture and radio-telemetry locations of each animal for the entire time we monitored it. On each day that we found at least one skunk dead from rabies (hereafter called death-day), we plotted the activity centers of all skunks captured to that day and identified all skunks with rabies when they died.

We tested the hypothesis that the observed spatial pattern of deaths among rabid skunks was random at each death-day, and across all death-days, by using a procedure described in Manly (1991). We computed the distance to the nearest neighbor (hereafter called nearest-neighbor distance) among all dead rabid skunks at each death-day and the mean of those distances. Then, keeping the location fixed of the first skunk to have died with rabies on the first death-day, we computed for each subsequent death-day a distribution of 1000 mean nearest-neighbor statistics by randomly assigning all dead rabid skunks among all known locations for all skunks to that day (beginning with death-day = 2). The number of skunks randomly-assigned at each death-day equaled the number observed to that date. Next, by averaging across all death-days, we computed the overall mean nearest-neighbor distance for the observed and the randomized distributions. To test the hypothesis, we computed the percentage of mean nearest-neighbor statistics from the randomized distribution that were equal to or greater than the observed mean nearest-neighbor statistic. We used the IML procedure of SAS (SAS Institute, Inc., 1989b) to conduct the randomization testing procedure.

We further evaluated relations among locations of skunks that died of rabies in 1992 with Mantel's test method (Manly, 1991). We tested the hypothesis that skunks whose activity centers were close together spatially would also have dates of death that were close together temporally. To do this, we computed for all skunks that died of rabies (found dead, immobile, and killed by predator) a spatial-distance matrix of all possible pairs of distances between activity centers and a time-distance matrix of all possible pairs of dates of deaths. We then computed a Pearson correlation coefficient (r) by pairing, on an element-by-element basis, the spatial-distance matrix and the time-distance matrix. We computed a randomized distribution of r-values by randomly assigning the spatial-distances to the time-distances 1000 times. Then, we computed the percentage of r-values from the random distribution that exceeded r-values from the observed distribution. We used the IML procedure of SAS (SAS Institute, Inc., 1989b) to conduct the randomization testing procedure.

We determined the number of times individual radio-collared skunks were observed at feeding sites during night tracking as a percentage of the total number of times they were relocated away from their daytime retreats. Mean values were computed, weighted by the square root of the number of relocations. We used an ANOVA to assess the relationship of sex, rabies, and their interaction to estimated percent of time in a night that skunks were found at feeding sites. We also determined if any skunk visited more than one feeding site during the monitoring period or if more than one skunk was at an individual feeding site at the same time.

Previous Section-- Introduction
Return to Contents
Next Section -- Results

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Webmaster
Page Last Modified: Saturday, 02-Feb-2013 05:57:40 EST
Sioux Falls, SD [sdww54]