ABSTRACT
Approximately 1.86 million baits containing a vaccinia-rabies glycoprotein recombinant vaccine were distributed with helicopters, vehicles, and bait stations during 2006–10. A bait density of 250 baits/km2 effectively controlled rabies cases in enzootic and preepizootic areas. However, a cluster of 11 rabid raccoons at the eastern edge of infection resulted in the initiation of semiannual, high-density (500 baits/km2) vaccination campaigns in approximately 20% of the oral rabies vaccination zone during July and September (2007–09). Bait success (i.e., chewed sachets or removed baits) at bait stations was negatively associated with station distances from water. Conversely, bait success improved with increasing distances from roads. Bait stations deployed significantly more baits in developed open space when compared to low- and medium- to high-intensity developed areas. However, a difference was not detected between developed open space and forest habitats. Rabies was confined to 86 raccoons within 317 km2 (10%) of a 3,133 km2 suburban landscape, with a disproportionate number of rabid raccoons (n=74) in developed areas, when compared to 10 cases in forest–wetland habitats. Two rabid raccoons did not fall within either general land-use classification. Rabies advanced 15.1 km eastward at a rate of 6.4 km/yr during a 28-mo interval (2004–06).
INTRODUCTION
Oral rabies vaccination (ORV) programs (Rupprecht et al. 1986) have been initiated in North America to reduce the incidence of rabies and to prevent continued viral advance through immunization of wildlife populations (Robbins et al. 1998; Boulanger et al. 2008; Slate et al. 2008; Sattler et al. 2009). Various control methods were used to eliminate rabies from Texas canids (Sidwa et al. 2005) and raccoons (Procyon lotor) in Ontario (Rosatte et al. 2007, 2008) and Quebec (Mainguy et al. 2012). Conversely, elimination of the raccoon rabies variant (RABV) has been problematic in the US (Slate et al. 2008).
A rabies epizootic erupted in the northeastern US following the translocation of infected raccoons to hunting camps in West Virginia during the 1980s (Jenkins and Winkler 1987). Raccoon rabies entered New York in 1990, with profound impacts to the public health system (Wyatt et. al. 1999; Chang et al. 2002). Two suburban counties on Long Island were of particular concern to the New York State Department of Health because high-density human and raccoon populations coexist in close proximity (Willsey et al. 2004). Terrestrial rabies was first reported in Nassau County in 1991; however, epizootic status was not observed until August 2004. The Ontario point infection control strategy (Rosatte et al. 2001) was modified and implemented in a 172 km2 area immediately following confirmation of the first rabid raccoon (Willsey et al. 2004), which included a target density of 120 baits/km2 (Recuenco et al. 2009). Both the extent of the affected area and number of cases increased, resulting in program transition to the New York State Veterinary Diagnostic Laboratory in November 2006. Vaccine deployment methodology and the landscape epidemiology of RABV in Nassau and Suffolk counties (NAS-SUF) are reviewed. Rabies virus neutralization results are presented elsewhere (Bigler et al. 2021).
MATERIALS AND METHODS
Study area
The suburban area (3,132.7 km2) is characterized by the coastal lowlands ecozone (Dickinson 1983) and bounded by New York City, the Long Island Sound, and the Atlantic Ocean (Fig. 1). The geologic foundation is comprised of Cretaceous sedimentary rock, with an overlay of relatively poor soils consisting of clays, sands, and gravels that are typically associated with glacial outwash. Approximately 2,832,882 people (US Census Bureau 2010) reside in NAS-SUF (40°46′N, 73°26′W). A 1-km radius was constructed around case locations (2004–09) to estimate 335,695 (i.e., 11.8% of 2,832,882) people in the 317.4 km2 polygon.
Rabies surveillance
Rabies surveillance during 2004–08 included routine public health submissions, as well as animals that were rabies-suspect (i.e., sick or behaving abnormally), found-dead, road-killed, or live-trapped by nuisance wildlife control or animal control officers. Enhanced rabies surveillance continued after 2008; however, submissions were restricted to rabies-suspect and found-dead animals.
Rabies diagnosis was performed with a direct fluorescent antibody assay (Centers for Disease Control and Prevention 2017). Variant-typing procedures (Szanto et al. 2008) confirmed RABV during 2004 and were discontinued until 2016.
Bait distribution
Fourteen ORV campaigns were conducted during 2006–10 (Fig. 2). Five September campaigns encompassed the entire ORV zone (2006–10), three included the eastern epizootic area during July (2007–09), four comprised bait-station applications during the spring (April-May: 2007–10), and two followed annual September campaigns (December 2008, October 2010.)
Fishmeal polymer baits (FMP) containing RABORAL V-RG (Boehringer Ingelheim Animal Health USA Inc., Athens, Georgia, USA) were distributed with bait stations (September 2006–07, April/May 2007–10), vehicles, and helicopters (2006–10). A 30-km, rabies-free barrier extended beyond the easternmost case during 2006–07. Rabies advance was not observed during 2007 and barrier depth was reduced to 20 km during 2008–10. The largest ORV area encompassed 1,321.8 km2 in September 2006; the smallest area (724 km2) was treated in September 2010. Clear sachets were provided in FMP during 2006–08; opaque white sachets were supplied in 2009–10.
A target density of 250 baits/km2 was established in September 2006. In July 2007, a contingency response campaign (CRC) was applied in response to a cluster of 11 epizootic cases during February-July (Fig. 1). Approximately 165–223 km2 of each autumn ORV zone was treated at a density of 500 baits/km2 during July 2007–09 and September 2007–10 (Fig. 2). Immediate CRC was also conducted in response to individual cases during December 2008 and October 2010.
Helicopter flight-lines were constructed in parallel or crosshatch configurations with 500-m intervals between lines. Geocoded waypoints were added to flight tracks to provide target quantities of baits that were manually deployed along linear intervals during real-time, GPS flight monitoring. A distribution rate of 5–6 baits/s was targeted to achieve a density of 250 baits/km2 at speeds between 74–93 km/hr. Aerial cell areas averaged 16.2 km2 and 32.2 km2 during July and September campaigns, respectively.
Greenspaces (e.g., cemeteries, parks) within ground cells were treated with vehicles or bait stations, depending on the assigned ground-baiting method in the area. Ground cells averaged 5.3 km2 during July and September campaigns. Achieved bait densities for all cells were calculated by dividing distributed bait totals by the cell areas.
County vehicle crews were provided with digital-orthoquad and street maps with a 1-km2 grid overlay. Previously described bait stations (Boulanger et al. 2006) were modified with 1.52 m lengths of schedule-40, polyvinyl chloride pipes, tee-joints, and end caps with a 10-cm diameter to accommodate up to 340 FMP. One or three lag bolts were affixed to both ends of the tee joints to discourage nontarget species interference. Each station was strapped to a tree or two iron reinforcing rods inserted in the ground, depending on the habitats at predefined GPS coordinates at staggered 2-km intervals within rows, with a 1-km distance between rows (Boulanger et al. 2006). Bait stations (n=138 and n=74) were distributed during September 2006 and 2007, respectively, replenished on day 10 if needed and removed from the field on day 20. The area of the autumn zone was reduced in 2007 and discontinued in 2008 following a decline in rabies cases.
Bait stations (n=77–98) were established in discrete areas of concern throughout the ORV zone following freeze-thaw cycles during 2007–10. Spring stations were checked at approximate 10-d intervals and removed 10–91 d following deployment, depending on bait disappearance rates. Baits were removed a minimum of 30 d prior to ORV. All FMP were counted, and bait status was recorded for spring and autumn baits. Bait success included chewed sachets or removed FMP. Intact baits and sachets were classified as bait failures.
Twelve motion-sensor cameras (Cuddeback Expert, Non Typical, Inc., Park Falls, Wisconsin, USA) were positioned at bait stations during a 20-d interval to provide an index of species visitations during September 2007. A 4-min interval between photographs was maintained to reduce the possibility of missing animals entirely or amassing multiple photographs of monomorphic individuals (Koerth and Kroll 2000; Boulanger et al. 2006).
Geospatial and statistical analyses
Geospatial applications were executed with ArcGIS 9 (Environmental Systems Research Institute, Redlands, California, USA) and Street Atlas 2006 (Delorme, Yarmouth, Maine, USA), and their subsequent software upgrades. Aerial photographs (US Department of Agriculture 2006) and upgraded data layers were used to construct and modify ground and aerial cells. Demographic features were obtained from Census 2010 Tiger files (US Census Bureau 2010) and human population data were derived from census block records. The 2006 US Geological Survey Land Use Land Cover Data (Fry et al. 2011) were used to estimate land-use parameters. Shapefiles containing landowner parcel areas were provided by the Nassau County Department of Planning (Mineola, New York) and the Suffolk County Real Property Tax Service Agency (Riverhead, New York). Chi-square analysis of observed and expected rabies cases was completed online with the QuickCalcs website (GraphPad 2011). Regression analyses were executed with NCSS Statistical Software (NCSS, LLC, Kaysville, Utah, USA).
RESULTS
Study area
Developed open space and low-intensity areas dominate the suburban landscape (Table 1) with elevations ranging between sea level and 122.2 m. Enzootic and county land use proportions were similar (r(14)=0.92; 95% confidence interval [CI]: 0.77–0.98; P<0.0001). Road density and parcel area (i.e., lot size) were summarized for bait distribution methods (Table 2) in the largest ORV zone (2006) to provide a standardized mechanism for comparisons to other suburban locations that might contemplate ORV. The mean landowner parcel area in the helicopter zone was 5,810 m2 (SE: 200). The vehicle zone (1,761 m2, SE: 39) was similar to the bait-station zone (1,381 m2, SE: 36). Mean road length was comparable in the helicopter (188 m, SE: 1.6), vehicle (134.9 m, SE: 0.5), and bait-station (125.6 m, SE: 0.6) zones. However, road densities in the vehicle (9.2 km/km2) and bait-station (11.3 km/km2) zones were approximately two times greater than the helicopter zone (5 km/km2).
Rabies surveillance
An opossum (Didelphis virginiana) in Nassau County was the first confirmation of terrestrial rabies on Long Island (1991); however, it was not included in our data because animal location and variant-typing results were not available. During August 2004–January 2009, RABV was identified in 2.7% (95% CI: 2.2–3.3, n=86) of 3,132 raccoons (Table 3). Rabies spillover was not reported in any other species.
In March 2016, RABV was confirmed in 0.05% (95% CI: 0–0.2, n=1) of 1,860 rabies-suspect raccoons tested between January 2009 and December 2017 (Table 3). Rabies was also identified in a dead river otter (Lontra canadensis) that was collected from a Suffolk County beach (December 2016) and a domestic kitten (Felis catus) in Nassau County (November 2017).
The majority (47.7%, 95% CI: 37.1–58.3, n=41/86) of enzootic, rabid raccoons were reported in developed open space, followed by 26.7% (95% CI: 17.3–36.1, n=23/86) in developed low-intensity habitat (Table 1). Individual land-use classifications were collapsed into two broad categories and 84 rabid raccoons were assigned to either developed or forest-wetland groups and compared to the number of cases that might be expected when adjusted for the area in each classification. Observed and expected values varied significantly (χ2(2,84) 12.8; P=0.0003), with 59 rabid raccoons expected in the developed classification (i.e., 74 observed), and 25 expected in the forest-wetland habitats (i.e., 10 observed).
Successive incremental distances from the index rabies case were regressed against month of laboratory diagnosis (Smith et al. 2005) to estimate the easterly rate of RABV spread on Long Island (Fig. 3). Rabies progressed 15.1 km, at a rate of 0.53 km/mo (6.4 km/yr) during a 28 mo interval (2004–06). Westward advance was not modeled because three rabid raccoons were previously confirmed in Queens County (1993–2003).
Bait distribution
Fourteen ORV campaigns resulted in the distribution of 1,864,272 baits between 2006–10 (Tables 4, 5). Helicopters deployed 50.3% (n=937,642 of the total) vehicles placed 37.2% (n=693,084), and bait stations dispensed 12.5% (n=233,546). During July CRC, the average achieved density in helicopter and vehicle cells was 497 and 477 baits/km2, respectively. In September, the average achieved density in vehicle (263 baits/km2) and helicopter (264 baits/km2) cells exceeded the target density of 250 baits/km2. Average achieved densities of 485 and 481 baits/km2 were attained in helicopter and vehicle CRC cells, respectively, during September campaigns. A bait application of 1,080 FMP was applied in a 1.1 km2 area (982 baits/km2) immediately following confirmation of a rabid raccoon in the CRC zone in December 2008. The western ORV boundary was extended southward by 91.5 km2 and 23,002 baits (251 baits/km2) in October 2010 because of a rabid raccoon in Queens County, located 325 m west of the Nassau/Queens County border.
On average, 191 baits were removed or consumed at autumn bait stations (Table 6). Bait success was positively associated with distance from road (0.1–278.9 m), with a significant linear increase of six removed or chewed baits for every 10 m a station was positioned away from roads. A negative linear response was associated with distance from water (0.9–3,332 m); bait success significantly decreased by five baits for every 100 m a station was placed away from water. When compared to developed open space, significantly fewer baits were removed in developed low and medium/high-intensity areas. Station placement in forested habitats resulted in the removal of an additional 38 baits; however, bait success in forested habitat was not significantly different from open-developed space. The number of lag bolts did not influence bait success.
During spring 2007–10, 155,086 baits were placed in bait stations; 67% (range: 49–83%, n=104,630 baits) were removed or chewed. During September 2006, 46,740 baits were deployed in 268.1 km2 and 16,128 baits (60 baits/km2) were removed or chewed at stations. The autumn zone was reduced to 140 km2 in 2007, with 9,610 bait successes (69 baits/km2) of 31,720 FMP deployed. Stations that performed well in September 2006, also did so during September 2007 (r(74)=0.71; 95% CI: 0.58–0.81; P<0.0001). During September applications, 32% of the baits were chewed or removed, in contrast to 67% during spring.
Bait station visitations
During a 20-d interval in September 2007, 1,498 photographs recorded 1,801 animals among 12 bait stations. Raccoons were photographed at all stations and yielded 71% (n=1,277) of 1,801 observations. Other photographs included 8% (n=153) cats, 7% (n=122) birds, 6% (n=103) opossums, 3% (n=46) red foxes (Vulpes vulpes), 2% (n=45) gray squirrels (Sciurus carolinensis), four small rodents, one dog (Canis lupus familiaris), one cottontail rabbit (Sylvilagus floridanus), and one chipmunk (Tamias striatus). Vertebrate species were not observed in 3% (n=48) of photographs and no people were photographed. Opossums were observed at nine of 12 bait stations, with 70.5% (95% CI: 61.7–79.3) bait interactions and a range of 1–59 photos/station. Cats were photographed at 11 of 12 bait stations, with a range of 1–38 appearances/station. Fifty photographs (32.7%, 95% CI: 25.2–40.1) showed cats passing by bait stations, and 103 photos (67.3%, 95% CI: 59.9–74.8) displayed feline interest in baits or bait stations.
DISCUSSION
This multimodal, rabies management approach serves as a case study in which preemptive strategies were developed and modified during the viral containment phase, with the ultimate goal of RABV elimination in a suburban environment. Over 1.86 million baits were deployed during 2006–10. The easternmost case was reported in 2006 and RABV was restricted to 10% of the NAS-SUF landscape. A control strategy with 120 baits/km2 was not successful during 2004–05 (Recuenco et al. 2009). Accordingly, the target density was increased to 250 baits/km2. Initiation of CRC (500 baits/km2) occurred in response to a focus of 11 rabid raccoons during February–July 2007; three additional cases were reported through January 2009. Serology results did not support continuation of biannual intervention (Bigler et al. 2021) and July CRC was discontinued after 2009. Two CRCs were executed immediately following individual cases that were confirmed after September ORV campaigns in 2008 and 2010, to promptly curtail viral transmission during later years when RABV elimination appeared imminent. The final ORV campaign was completed in 2010.
The significance of bait success relative to bait station placement near water and within natural environments was anticipated because of raccoon habitat preferences (Gehrt 2003). However, bait success also improved when station distance from roads increased. Similarly, scent stations located in interior habitats evidenced greater visitation rates when compared to scent stations near roadsides (Helon et al. 2002). By its very nature, vehicle distribution along roadsides appeared to select against raccoon-bait interactions in suburban environments.
There was no difference in bait success when one or three lag bolts were affixed to bait stations. Bait accumulation at the base of stations provided ample opportunities for nontarget species interference. Exploration of restrictive designs (Smyser et al. 2015) and novel bait stations might improve effectiveness during future applications.
The categorization of autumn bait stations within different habitat classifications resulted in large standard errors when bait success was evaluated. However, when compared to developed open space, significantly fewer FMP were removed or chewed in developed low-, medium-, or high-intensity areas. Conversely, bait success was comparable in developed open space and forested habitats. Bait disappearance in a rural forest was also significantly greater, when compared to forest-edge and field transects (Boyer et al. 2011). Larger suburban forested areas are readily baited with helicopters. However, mowed expanses are often deliberately excluded from ORV because cover vegetation is minimal or absent, even though raccoons use mature trees, water, and food sources within manicured parks and cemeteries. The continued exploration of alternative distribution methods to facilitate strategic bait deployment within critical open spaces will serve to enhance ORV success, while possibly decreasing program costs with dispersal of fewer baits.
Distribution bias was likely associated with greater bait success during the spring when compared to autumn. Spring stations were typically placed in ideal raccoon habitats. Conversely, autumn stations were positioned in a uniform, grid pattern within a defined area, resulting in station placement in highly developed habitats with poor bait success. Stations were also limited to 20 d in autumn, whereas productive spring stations were refilled at 10-d intervals up to 91 d. Vaccine titers remained stable in bait stations during 30 d in full sun or shade.
Anthony et al. (1990) reported that urban raccoons were more likely to be live-trapped in single-unit residential areas when compared to multiunit residential zones. Developed-low intensity areas had the greatest likelihood of rabies infection when compared to forested habitats (Recuenco et al. 2008). Given the suburban habitats and elevated human density in NAS-SUF, it is not surprising that more rabid raccoons were observed in developed areas with preferred habitats. Conversely, rabid raccoons were reported less frequently in forest/wetland habitats, likely reflecting a human submission bias. When compared to more-developed habitats, the importance of developed open space and forest/wetland habitats was underscored by significantly greater raccoon seroconversion (Bigler et al. 2021) and success at bait stations. Depending on the influences of waterways and other landscape features in urbanized environments, a habitat-based intervention focusing primarily on natural areas might be as effective as uniform bait deployment, while serving to reduce vaccine and labor costs through exclusion of habitats that are less desirable to raccoons.
An average advancement rate of 38.4 km/yr was reported during the spread of the mid-Atlantic rabies epizootic, followed by a slower wave of 9.5 km/yr during the enzootic phase (Biek et al. 2007). The epizootic rate of 15 km/yr in the Cape May ORV zone (Roscoe et al. 1998) was most comparable to NAS-SUF, where rabies progression approached 6.4 km/yr while point-source control measures were in progress. Intervention within 24 h of initial case identification (Willsey et al. 2004) likely reduced intraspecific viral transmission and impeded easterly spread until broad-scale ORV was applied in 2006.
Raccoons were recorded in 90% of 522 photographs during a pilot bait-station investigation in New York (Boulanger et al. 2006). April raccoon visitation rates of 74.8% and 82.6% in Massachusetts and Florida, respectively (Bjorkland et al. 2017), were similar to the September observations (71%) in NAS-SUF. Conversely, 17.4% and 25.2% of the visits were attributed to opossums, compared to only 6% on Long Island. Unlike other locations, feline interference exceeded opossum observations in NAS-SUF.
Epidemiological investigations led by state and county health departments determined that a rabid raccoon, river otter, and kitten originated from enzootic locations outside of NAS-SUF during 2016–17 (Brunt et al. 2020). The kitten was adopted from a county north of New York City and confined indoors following translocation to Nassau County. Results derived from whole genome sequencing and phylogenetic analyses indicated that the virus samples from the live-trapped raccoon and found-dead river otter represented separate incursions and were related to RABV isolates in Connecticut. Conversely, RABV samples that were collected during the NAS-SUF epizootic (2004–09) were associated with virus isolates from southern New York and northern New Jersey.
Continued enhanced rabies surveillance is crucial to maintenance of rabies-free status. Areas are recognized as rabies-free when no cases are reported during a 2-yr interval (World Health Organization 2013). Rabies virus was not identified in NAS-SUF during 7 yr between January 2009–March 2016. Four years have also elapsed since March 2016, when one raccoon tested positive 8 yr after the last Nassau County case was reported in 2007. The juvenile male was likely translocated from Connecticut by means of accidental or purposeful human assistance, or it might have dispersed naturally through a substantial barrier consisting of New York City, a substantial river and tidal estuary, and possibly a major vehicle connection (e.g., bridge) to Long Island.
Nine of 10 Nassau County RABV cases were confirmed during August–September 2004. Conversely, adverse northern meteorological conditions might have inhibited an epizootic when a solitary case was detected in Nassau County during March 2016. A canine distemper virus outbreak was also confirmed in the same area as the translocated, rabid raccoon in 2016 (A. Davis pers. comm.). The adverse influences of season and distemper mortality might have contributed to insufficient interactions among conspecifics and prevented RABV transmission. Nonetheless, RABV detection in three nonresident animals during 2016–17 underscores the advantage and practicality of widespread ORV applications over extensive landscapes to reduce the potential of natural and human-assisted translocation events that jeopardize costly ORV programs and rabies-free areas. Accordingly, an ORV program was initiated (2014) in response to persisting RABV cases in juxtaposed portions of New York City, with the goals of RABV elimination and reducing the potential of reinfection in NAS-SUF.
Skunks (Mephitis mephitis) can serve as a temporal reservoir of RABV (Guerra et al. 2003), thereby posing substantial challenges to ORV success (Tolson et al. 1987; Grosenbaugh et al. 2007). Rabid skunks have never been reported in NAS-SUF. Furthermore, skunks were not observed, live-trapped, or found dead on roads during extensive field activities. Successful rabies elimination might be attributed, at least in part, to the absence of skunks during ORV intervention.
The ORV management tactics that were implemented during 2006–10 are not sustainable over the coastal extent of RABV infection between Maine and Florida. However, complex environments with elevated mammalian densities might require intensive intervention following ORV barrier breaches or during rabies-elimination campaigns, rather than permitting continued epizootic advance or unduly prolonging ineffective control measures. Aggressive ORV management strategies were ultimately cost-effective in NAS-SUF, with a net benefit to New York State and local jurisdictions exceeding $19.8 million in 2018 (Elser et al. 2016). Pursuits of novel and improved vaccines, baits and attractants should be continued, along with exploration of adaptive ORV strategies that decrease program costs, while enhancing efficiency, safety, and effectiveness during large-scale RABV elimination efforts in the US.
ACKNOWLEDGMENTS
The ORV program was supported by the New York State legislature and the Nassau and Suffolk County legislatures (2006–10). R. Formosa helped to coordinate field operations. County and state personnel (B. Gibbins, R. Pescatore, R. Jean, S. Mikuljan, B. Laniewicz, A. Willsey, and D. Bruno) and scores of others provided invaluable in-kind logistical support. In-kind aerial support was contributed by the Suffolk County Police, New York State Police, and Nassau County Police Marine Aviation Units. We are indebted to A. Davis and her colleagues for providing confirmation of canine distemper virus infection, as well as the development and completion of whole-genome sequencing and phylogenetic techniques to determine the origins of the rabies virus isolates that were identified in a river otter, kitten, and raccoon (2016–17). Virus titrations of the RABORAL V-RG vaccine following fishmeal polymer bait exposure to environmental conditions were performed by N. Zylich and E. Dubovi. Locations for rabid animals in New York City were provided by S. Slavinski.