Browsing by Subject "piecewise regression"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Evaluating antipredator strategies and ecological mechanisms that affect neonatal white-tailed deer (odocoileus virginianus) predation risk in Minnesota’s agricultural prairies
    (2024-12) Obermoller, Tyler
    Neonatal survival is a key driver of annual variation in ungulate population growth. Neonatal ungulates are vulnerable to many factors, but predation is often the leading cause of mortality. A key logistical challenge in studying neonatal survival is locating and collaring individuals, prompting us to evaluate new technologies to meet our study objectives. We found the use of a thermal infrared (TIR) equipped drone to be an efficient method under good conditions (e.g., cloudy, nighttime) to locate fawns in open habitats and outperformed other capture methods (e.g., vaginal implant transmitters, ground searches, monitoring doe behavior). Drones show immense promise in the future of wildlife research. Recent advances in biotelemetry enable monitoring of large mammals with varying neck sizes across multiple ecological milestones (e.g., summer survival, dispersal, winter habitat use). Radiocollars are instrumental in tracking home range size, dispersals, migrations, resource use, behavior, and causes of mortality. However, we identified that collar failures (e.g., broken, premature expansion, fence-related) often occurred after 6 months of age, which may reduce sample sizes necessary to examine dispersal, migration, and first-year survival. These findings emphasize the importance of aligning study design with specific ecological questions (e.g., summer survival, migration) and accounting for potential retention issues (e.g., barbed-wire fences, thick cover) when determining sample size. We deployed expandable global positioning system (GPS) collars on neonatal white-tailed deer (Odocoileus virginianus) fawns and modeled their activity patterns using piecewise regression models to determine more refined timing of antipredator strategies and how transitions changed with age by individual fawns. On average, fawns ended their hider strategy at 10.9 days old and entered the transitioner strategy for a duration of 25.5 days, and then initiated the follower strategy at 37.0 days old. Males were 1.8x more likely to die from predation than females. We also found fawns in the hider strategy had a 4.9x higher risk of mortality than those in transitioner strategy. These findings suggest that as fawns increase mobility via neuromuscular development, they shift from hiding to fleeing behaviors. This increased mobility enhances antipredator strategies, with more “fidgety fawns” demonstrating improved predator evasion tactics. We hope future researchers use our framework to examine other neonatal ungulates to further develop a robust antipredator ethogram (e.g., hider, transitioner, follower) with varying predator cohorts to examine interspecific ungulate variation that will yield insight into neonatal risk and ungulate population dynamics. Evaluating bedsite selection is a key component of neonatal risk and predator avoidance. Fawns are most susceptible to predation early in life and better understanding predisposing factors may be important to juvenile recruitment. Therefore, proper bedsite selection to avoid predator detection likely plays a critical role in survival. We compared mortality sites of fawns predated by coyotes with paired and capture bedsites of surviving fawns to assess variation in habitat characteristics that may be important to predation risk. We also compared bedsite characteristics by neonatal antipredator strategy (e.g., hider, follower) to further examine variation in selection. Our results revealed no differences with fawn sightability or lower canopy cover at paired and capture bedsites compared to mortality sites. We also found no differences with concealment cover, suggesting that concealment was adequate throughout the study area. Finally, we investigated habitat selection of fawns by antipredator strategies (e.g., hider, transitioner, follower) using an integrated step selection analysis (iSSA), and evaluated 3 predator reduction hypotheses: prey hiding hypothesis (PHH, selection for habitats with increased cover), predator escape hypothesis (PEH, selection for habitats with increased visibility), and landscape of fear hypothesis (spatial avoidance of predators, LFH). We found distinct differences in habitat selection across strategies, with hiders and transitioners favoring cover-rich habitats like grasslands and forests, aligning with the PHH. In contrast, followers exhibited behaviors consistent with the LFH by avoiding coyote travel corridors, such as roads and croplands. These insights provide a deeper understanding of the behavioral adaptations neonatal ungulates use to reduce predation risk and contribute to a broader understanding of predator-prey dynamics.
  • Thumbnail Image
    Item
    A multi-faceted evaluation of a reintroduced waterfowl species: Migration ecology, ecotoxicology, and population genetics of trumpeter swans in the Midwest
    (2024-06) Wolfson, David
    Trumpeter swans (Cygnus buccinator), the largest waterfowl species in North America, were widespread throughout much of the continent prior to European colonization. Due to unregulated market hunting, trumpeter swans were nearly extirpated, and reached an estimated low of ~70 individuals in the lower 48 U.S. states during the 1930s. The creation of new protected areas coupled with management efforts allowed trumpeter swan abundance to recover, and they were successfully reintroduced to the western Great Lakes region in the late 20th century to re-establish the Interior Population (IP). However, a general lack of information about IP trumpeter swan ecology has hindered conservation decision-making. I partnered with agency biologists from seven U.S states and one Canadian province to gather information on IP trumpeter swan annual movements and migration patterns, the prevalence of lead (Pb) in free roaming swans, and the genetic makeup of trumpeter swan populations in North America. Before addressing the previously mentioned aspects of trumpeter swan ecology, in Chapter 1, I provide a review of piecewise regression, a flexible type of breakpoint analysis. I provide an overview of piecewise regression and then describe six case-studies, using piecewise regression on a variety of datasets that include a range of species, data types, ecological responses, statistical signals, and timeframes. In Chapter 2, I use the methodology described in Chapter 1 (i.e., piecewise regression) to quantify annual movements and migration patterns in IP trumpeter swans. We (multiple state agency biologists and other collaborators) deployed 133 GPS-GSM transmitters on trumpeter swans across the current IP breeding range (i.e., the greater Midwest) during 2019–2022. Individual tracking data revealed that IP trumpeter swans are partial migrants, with a continuum of strategies each year, from local movements to long-distance migration. Much of the variability in movement patterns was related to factors tied to natural history demands (e.g., breeding status) and response to environmental conditions (e.g., through associations with breeding latitude). In Chapter 3, I present a baseline assessment of the prevalence of lead in all trumpeter swans associated with the dissertation and an additional flock in Nebraska. I estimated blood lead concentration for 119 IP trumpeter swans and detected lead in all individuals. However, 91% of swans had blood lead levels in the ‘background’ range (not considered to produce negative physiological effects), 7.5% of swans had blood lead levels in the ‘sub-clinical’ range, and only 1.5% of swans had blood lead levels in the ‘clinical’ or ‘severe’ range of lead toxicity. Finally, in Chapter 4 I present a comparative assessment of the genetics of trumpeter swans in North America. I collected 150 genetic samples from IP trumpeter swans captured during 2019–2022 and also obtained 79 reference samples from the other two North American trumpeter swan populations. These samples provide evidence that all three populations are genetically distinct and that the High Plains flock of the IP has lower genetic diversity compared to the other groups, likely a result of smaller population size, relative geographic isolation, and potential founder effects.