Reproductive success, habitat selection, and neonatal mule deer mortality in a natural gas development area

Abstract

Mule deer (Odocoileus hemionus) populations have periodically declined throughout the western United States, with notable declines during the late 1960s, early 1970s, and 1990s (Unsworth et al. 1999) to present. Declining population levels can be attributed to low fawn survival and subsequently low population recruitment (Unsworth et al. 1999, Pojar and Bowden 2004) caused by declining habitat availability and quality (Gill 2001, Lutz et al. 2003, Watkins et al. 2007, Bergman et al. 2015). Although, general public perception is that declining deer numbers are attributed exclusively to predation (Barsness 1998, Willoughby 2012), predator control research suggests otherwise (Hurley et al. 2011, Kilgo et al. 2014) and compelling evidence exists that improving habitat quality can enhance deer populations (Bishop et al. 2009, Bergman et al. 2014). Complicating this story is the large-scale habitat alterations driven by natural gas development, which may also influence deer population dynamics. Natural gas development and associated disturbances that can affect deer habitat and population dynamics include conversion of native plant communities to drill pads, roads, or noxious weeds and noise pollution from compressor stations, drilling rigs, increased traffic, and year round occurrence of human activities. Natural gas development alters mule deer habitat selection through direct and indirect habitat loss (Sawyer et al. 2006, Sawyer et al. 2009, Northrup et al. 2015). Direct habitat loss results from construction of well pads, access roads, compressor stations, pipelines, and transmission lines. Activity, traffic, and noise associated with increased human presence and development may lead to indirect habitat loss. Indirect habitat loss is exacerbated because active wells produce gas for 40 years or longer (Sawyer et al. 2006, Sawyer et al. 2009). In addition, indirect habitat loss affects considerably larger areas than direct habitat loss (Sawyer et al. 2006, Sawyer et al. 2009). Recent research suggests direct and indirect losses can lead to behavioral responses to development (Sawyer et al. 2006, Dzialak et al. 2011b, Northrup et al. 2015). However, deer can behaviorally mediate these impacts by altering activity patterns or selecting habitat with topographic diversity that provides refuge from development (Northrup et al. 2015). Obtaining a more complete understanding of the potential impacts of development is critical to comprehend population dynamics of deer and to develop viable mitigation options. Understanding how natural gas development and other factors influence reproductive success metrics (e.g., pregnancy, in utero fetal, and fetal survival rates), fetal sex ratio, habitat characteristics of birth and predation sites (i.e., habitat selection), and neonatal (i.e., 0–6 months old) mule deer mortality have been identified as knowledge gaps. Thus, my dissertation focused on addressing these knowledge gaps through individual reproductive success monitoring using vaginal implant transmitters. I conducted this research during 2012–2014 in the Piceance Basin of northwestern Colorado in study areas with relatively high (0.04–0.90 well pads/km2) or low (0.00–0.10 well pads/km2) levels of natural gas development. In chapter 1, I examined the influence of adult female, natural gas development, and temporal factors on reproductive success metrics (i.e., pregnancy rate, in utero fetal rate, and fetal survival rate) and fetal sex ratio. Pregnancy rates were high, did not vary across years, and were essentially equal between study areas. In utero fetal rates were lower for yearling females (n = 10) and varied annually compared to adult females (n = 204) possibly from annual weather patterns that influenced forage quality and digestibility. Fetal survival rates increased over time and were lower in the high development study areas than the low development study area in 2012 possibly caused by a compounding influence of development disturbance during extreme environmental conditions (i.e., drought). Higher road density in a female’s core area (i.e., 50% minimum convex polygon) on summer range possibly contributed to better maternal body condition through increased forage quality along roads. Following the Trivers-Willard hypothesis which predicts females in good versus poor condition will produce more males, my results suggested females had a higher probability of producing more male offspring as road density increased. However, under my proposed mechanism, I would expect body condition and road density to be strongly correlated, but they were only weakly correlated (r = 0.07). I also note that I did not detect a biased sex ratio at the population level. Thus, I am uncertain of the exact mechanism influencing the relationship between road density and fetal sex ratio. In chapter 2, I used global positioning system collar data in conjunction with VITs and linear mixed models to validate the use of maternal deer movement rates (m/day) to determine timing of parturition. Daily movement rate of maternal deer decreased by 39% from 1 day before parturition ((x ) ̅ = 1,243.56, SD = 1,043.03) to 1 day after parturition ((x ) ̅ = 805.30, SD = 652.91). Thus, I suggest that a mule deer female whose daily movement rate significantly decreases to ≤ 800 m/day has likely given birth. In the future, I will analyze an independent data set to validate the recommended threshold value and possibly develop a movement rate algorithm. In chapter 3, I fit resource selection functions to examine the influence of natural gas development and environmental factors on birth site selection and habitat characteristics of predation sites. Females selected birth sites farther from producing well pads and with increased cover for concealing neonates and appeared to select habitat (e.g., north-facing slopes and further from treed edges) that minimized neonate predation risk. Predation sites were characterized as being closer to development and in habitat (e.g., woodlands, aspen-conifer stands, and north-facing slopes) that possibly provided favorable microclimates for neonates and abundant high quality forage for lactating females. However, I note that predation sites were on average relatively far (2,057 m) from producing well pads and I have difficulty proposing a mechanism to explain how well pads that far away can influence predation site characteristics. My results suggest natural gas development and environmental factors (e.g., slope, habitat type, and aspect) can influence birth site selection with predation site characteristics possibly related to foraging habitat selection. In chapter 4, I tested hypotheses about the influence of adult female, natural gas development, neonate, and temporal factors on neonatal mortality using a multi-state model. Predation and death by malnutrition decreased from 0–14 days old. Predation of neonates was positively correlated with rump fat thickness of adult females, but negatively correlated with the distance (0–0.4 km) from a female’s core area (i.e., 50% kernel density estimate) to a producing well pad on winter or summer range. Death by malnutrition was positively correlated with the distance from a female’s core area to a road on winter range and weakly, but negatively correlated with temperature. During my study, predation was the leading cause of neonatal mortality in both areas and mean daily predation probability was 9% higher in the high versus low development areas. However, black bear (Ursus americanus) predation was the leading cause of neonatal mortality in the high development areas (22% of all mortalities) compared to cougar (Felis concolor) predation in the low development areas (36% of all mortalities). Reduced precipitation and patchy habitat further fragmented by development possibly contributed to less hiding cover or edge effects, potentially leading to increased predation in the high development areas. Overall, my results suggest natural gas development may decrease fetal survival, influence birth site selection, and increase neonatal mortality, especially through predation, which may have consequences for mule deer recruitment and population dynamics depending on development intensity, habitat, and environmental conditions (e.g., drought). Consequently, developers and managers should consider strategies to mitigate impacts from development and improve forage and habitat quality and availability to minimize fitness consequences of deer. Such strategies could include development planning to avoid important habitats during critical time periods, implementing habitat treatments to rehabilitate areas, and minimizing habitat fragmentation and removal of hiding cover when constructing well pads and roads.