Skip to main content

Integrating Multiple Databases to Evaluate the Impact of Weather on Animal Movements

(Under the direction of Drs. Suzanne Kennedy-Stoskopf, Christopher DePerno, Alun Lloyd, and Stacy Nelson)

Wildlife ecologists strive to correctly model and predict complex relationships between animals and their environment and facilitate management decisions of delicate ecosystems. Spatiotemporal patterns of wildlife provide an interface for understanding movements and interactions with their environment. Recent advances in monitoring systems across scientific domains have made it possible, yet challenging, to combine and use detailed output from various sources to address questions spanning multiple disciplines.

To address these complex questions, WolfScout wildlife and weather tracking system was designed to automate integration of the latest technological advances in GPS collars, weather stations, drought conditions, severe weather reports, and animal demographic information (Chapter 1). Also, WolfScout spatial database management system stores a variety of landscape maps and calculates distance between animal locations and landscape characteristics. The distance data, environmental conditions, and animal information are fully downloadable through a secure website. WolfScout was designed to promote interoperability of data between researchers and software applications while standardizing analyses of animal interactions with their environment. However, analyses of this intimate view of wildlife and the environment presented problems from a lack of statistical methods that handle “big data” questions. To overcome these challenges, a recursive partitioning method (i.e., random forest) was used to identify variables of importance and condense datasets to manageablesizes for traditional regression methods.

Lauren Elizabeth Charles-Smith

Lauren Elizabeth Charles-Smith

To demonstrate WolfScout’s capabilities, GPS locations from eastern coyote (Canis latrans; Chapter 2) and white-tailed deer (Odocoileus virginianus; Chapter 3) residing in the Sandhills region of North Carolina were streamed into the system and processed. From WolfScout, distances to the closest water centerline, all available variables from closest weather stations, multi-sensor precipitation estimates, vapor pressure deficits (VPD), and drought conditions per animal location were downloaded. The centerlines provided the only areas of dense cover in the longleaf pine-wiregrass ecosystem, otherwise characterized by porous, sandy soils and sparse vegetation. For each species, variables of importance were identified with random forest methods and stepwise Akaike information criterion and variance inflation factors were calculated to identify top regression models of weather influence on distance to lowlands. Both species were relocated in bottomlands during day and early evening hours. The animals were closest to the water centerlines during denning/breeding for coyotes and late gestation/parturition for deer. Increasing VPD and dry conditions correlated with lowlands for coyotes and deer. Coyotes were in lowlands during high winds, low pressure, and low temperatures with high relative humidity and dew point temperatures. In winter, deer were located in areas sheltered from northeasterly high winds at times of high visibility, pressure, temperature, and humidity. In spring and summer, deer were in lowlands during precipitation, increased chance of precipitation, and impending storms.

To complement these results, the influence of severe weather events, weather station variables and drought conditions on human-mesocarnivore interactions in the Sandhills area were analyzed (Chapter 4). In general, most interactions occurred during severe drought and high VPD. Specifically, non-rabid fox encounters were at times of high VPD and wind gusts, whereas human-rabid fox interactions occurred during moderate droughts. High temperatures, relative humidity, precipitation and wind speeds from the southwest, and parturition time characterized non-rabid raccoon encounters. Human-rabid raccoon interactions occurred during severe weather events, high pressures, high dew point temperatures and high wind speeds. In general, fox were encountered during dry conditions and raccoons were encountered during times of severe weather and precipitation

Through the design and implementation of WolfScout wildlife and weather tracking system, climatic variables have shown to influence wildlife locations, including encounters with humans. This weather information should be included in subsequent analyses aiming to predict the spatial-temporal dynamics of wildlife.


Henry Jared Flowers

Henry Jared Flowers


Methods for Monitoring and Assessing Sturgeon Populations Using Technology.

(Under the direction of Joseph E. Hightower.)

Sturgeons (Acipenseridae) are one of the most threatened taxa worldwide, including populations of Atlantic Sturgeon Acipenser oxyrinchus oxyrinchus. This subspecies is found in rivers and marine areas of the Atlantic Coast of the United States. The National Oceanic and Atmospheric Administration listed Atlantic Sturgeon under the Endangered Species Act in 2012. The listing delineated five Distinct Population Segments (DPS), one classified as threatened (Gulf of Maine) and five as endangered (New York Bight, Chesapeake Bay, Carolina, and South Atlantic). Populations are significantly reduced from historic levels by a combination of intense fishing and habitat loss. Successful restoration of Atlantic Sturgeon depends on a solid foundation of biological data.

Traditional fisheries sampling approaches are often limited for federally-listed species such as Atlantic Sturgeon. One alternative is to develop less-intrusive methods using new technologies. Side-scan sonar is an emerging fisheries technology with advantages over traditional sampling techniques, including the ability to efficiently sample large areas and survey fish without physically handling them – important for species of conservation concern. Acoustic telemetry is another area of rapid technological advances, including smaller, longer-lived tags and new designs for submersible receivers. Online databases facilitate data-sharing between researchers operating autonomous receiver arrays, enabling telemetry studies to incorporate expansive spatial areas.

The first objective for this study was to develop methodology using side-scan sonar to survey and assess Atlantic Sturgeon populations. This was accomplished by surveying six North and South Carolina rivers, using a combination of side-scan sonar, telemetry, and video cameras (to sample jumping sturgeon). We surveyed lower reaches of each river, near the saltwater/freshwater interface, on three occasions (generally successive days) and used occupancy modeling to analyze these data. We were able to detect sturgeon in five of six rivers, with estimated gear-specific detection probabilities ranging from 0.2-0.5 and river-specific occupancy estimates (per 2-km river segment) ranging from 0.0-0.8.

Next we used count data from the same side-scan sonar surveys to estimate abundances of sturgeon >1 m in length in conjunction with N-mixture and distance models. Estimated abundances in the Carolina DPS were 2,031 (95% confidence interval: 1,075-3,858) and 1,912 (1,016-3,616) using N-mixture and distance models, respectively. The Pee Dee River, South Carolina had the highest overall abundance of any river at 1,944 (1,036-3,646) and 1,823 (976-3,406) using count and distance models, respectively. These estimates do not account for sturgeon occurring in unsurveyed riverine reaches or marine waters. Comparing the two models, the N-mixture model produced similar estimates using less data than the distance model with only a slight reduction of estimated precision.

The third objective was to use telemetry to describe long-term movements of Atlantic Sturgeon from the Roanoke River, North Carolina. Six adult Atlantic Sturgeon (presumably males) were implanted with acoustic telemetry tags from 2010-2012. Sturgeon were monitored through a network of passive receivers in North Carolina and eight additional states. A multi-state model was used to estimate movement probabilities among riverine, estuarine, and marine areas. From September 2010 to June 2014, five of six Atlantic Sturgeon were detected in marine portions of three different DPSs. Seasonally, sturgeon were observed to either spend the entire year in marine waters or winter-spring in marine waters, summer in Albemarle Sound and fall in the Roanoke River for spawning. The multi-state model suggests seasonally variable movement probabilities. Estimated annual mortality was relatively low (0.03) and detection probability high (>0.50) in most study regions. Sturgeon were observed to spawn in consecutive years or with a year in between spawning events. The complexity of Atlantic Sturgeon movements and the mixing of populations in marine waters add to the potential difficulty in managing the recovery of this species.


Timothy Alan Ellis

Timothy Alan Ellis


Mortality and Movement of Spotted Seatrout at Its Northern Latitudinal Limits.

(Under the direction of Drs. Jeffrey A. Buckel and Joseph E. Hightower.)

Spotted seatrout (Cynoscion nebulosus) is one of the most economically important sportfish in the U.S. South Atlantic and Gulf of Mexico. The species is uncommon north of Chesapeake Bay but supports large recreational and commercial fisheries in North Carolina and Virginia. However, despite its importance, relatively little is known about spotted seatrout mortality and movement at the species’ northern latitudinal limits, including winterkill and the appropriate spatial scale for management. Data from the first comprehensive tag-return study (i.e., high-reward and double tagging) of spotted seatrout in North Carolina were used to estimate bimonthly rates of fishing mortality (F) and natural mortality (M). From September 2008 to October 2012, 6,582 spotted seatrout were tagged and 557 (8.5%) tags were recovered and reported by fishers through August 2014. Estimates of bimonthly F and M varied seasonally and ranged from 0.024-0.139 and 0.062-2.527, respectively. Contrary to findings from North Carolina’s recent age-based stock assessment, annual mortality from natural causes was always identified to be much higher than mortality from fishing. In a concurrent telemetry study, spotted seatrout in two North Carolina estuaries were tagged with acoustic transmitters and monitored during three consecutive winters of varying severity. Fates of telemetered fish were inferred from daily movements and used in a multistate capture-recapture model to estimate weekly F, M, and emigration rate. Again, fishing mortality was found to be low relative to natural mortality during most weeks with few inferred harvests of telemetered fish (weekly estimates of F ranged from 0.012-0.078) and more natural deaths (weekly M ranged from 0.012-0.300); most natural deaths occurred when water temperature was below 5 °C. Field estimates of M were supported by the results of two laboratory experiments on the cold tolerance of spotted seatrout, as affected by rapid and prolonged exposure to low-temperature extremes (3, 5, or 7 °C) across upper- (10 ‰) and lower-estuarine (30 ‰) salinities. These empirical estimates of cold tolerance were used to develop temperature-based models for predicting winter M, which were then applied to 1994 to 2013 water temperature data. Estimates of winter M from the last 20 years that were high matched historical winterkill events. Lastly, tag recovery information was used to describe the large-scale movements of spotted seatrout in North Carolina. Regional differences in movement suggest that most spotted seatrout reside year-round within the major estuarine systems of North Carolina but that fish tagged in the northern Outer Banks region tend to be transient. For reported recaptures of spotted seatrout tagged in North Carolina (n = 499), 49 (9.8%) were from outside of the state’s jurisdictional boundaries, primarily Chesapeake Bay. Extensive movements southward during fall months and northward during spring and summer months are likely associated with known overwintering and spawning periods for spotted seatrout. In combination, these studies provide much new information about spotted seatrout ecology, stock boundaries, and the relative importance of fishing and natural morality, which will improve future management of this species throughout its geographic range.