Population Monitoring

Contributed Oral Presentations

SESSION NUMBER: 45

Contributed paper sessions will be available on-demand for the duration of the conference, then again at the conclusion of the conference.

 

Estimating Encounter Location Distributions from Animal Tracking Data
Michael J. Noonan; Ricardo Martinez-Garcia; Christen H. Fleming; Justin M. Calabrese
Linking individual behavior with higher-level processes has long been a cornerstone of ecological research. For motile species, this `upscaling’ is governed by how well any given movement strategy maximizes encounters with positive factors (e.g., food, mates, essential resources), and minimizes encounters with negative factors (e.g., predators, and disease). Despite the importance of encounter events for a broad range of ecological processes, encounter theory has not kept pace with the developments in animal tracking or movement modeling. Furthermore, existing work has focused primarily on the relationship between animal movement and encounter rates while no study has directly related individual movement with the spatial locations of encounter events in the environment.
Here, we bridge this gap by developing a formal framework for estimating a spatial distribution of encounter probabilities, termed the conditional distribution of encounters (CDE). Using simulated data, we show how our method provides asymptotically consistent estimates. We then demonstrate the general utility of this method for three simulation-based scenarios that occur routinely in biological systems: i) a population of individuals that occupy relatively distinct home ranges, but with some degree of spatial overlap with neighbors; ii) a pair of individuals with a hard territorial border between their home ranges; and iii) a predator with a large home range that encompassed the home ranges of multiple prey species. We also show here how ridge estimation on the CDE can provide an accurate, and objective estimate of territorial borders. We then demonstrate the real-world utility of the CDE on GPS data from white-faced capuchins (Cebus capucinus) tracked on Barro Colorado Island, Panama, and sleepy lizards (Tiliqua rugosa) tracked in Bundey, South Australia. We conclude with a discussion of the potential applications of this novel target distribution for better understanding the dynamics of populations of interacting individuals.
Efficient Population-Level Animal Movement Modeling in Continuous Time
Christen H. Fleming; Daniel Sheldon; Jonathan Dresher-Lehman; Michael J. Noonan; Tom Akre; William F. Fagan; Justin M. Calabrese
Continuous-time movement models have many advantages, such as the capability to handle irregularly timed observations and erroneous location estimates, while producing biologically interpretable parameters and avoiding the problem of scale dependence. Related population-level inferences have traditionally been performed by averaging point estimates or estimating hierarchical models via MCMC methods. The former practice is a crude approximation to meta-analysis, because it ignores parameter uncertainties, while the latter practice can be computationally prohibitive. Here, we introduce a complementary meta-analysis and hierarchical modeling framework for continuous-time movement models that correctly propagate uncertainties and are computationally efficient. We apply this framework to multiple example species, including South American jaguar (Panthera onca) and wood turtle (Glyptemys insculpta), to show how appropriate methods can be critical to biological inference.
Beyond Your Ears: VHF Tag Detection via Coherent Pulse Integration
Michael W. Shafer; Paul Flikkema
Very high frequency (VHF) telemetry tags are widely used on species too small to carry other transmitter technologies, as well as on larger species who might carry an archival tag for data collection and a VHF transmitter for localization prior to recapture. Our group has developed an open-source UAV radio telemetry system (UAV-RT) for VHF radio tag localization that combines wideband software defined radios (SDR), automated flight controllers, and custom data processing software. The 3D mobility of the UAV-RT system allows for rapid movement in the search area, improved vantages, and the SDR enables novel signal processing techniques. In the process of developing this system, we have found that reliable VHF signal detection is the core challenge for wildlife radio telemetry, and is a fundamental constraint to current and future localization methods, both on the ground and aerial, but requires a systemic approach for true improvement. We define detection as the positive decision that a tag’s signal has been received. This detection represents a form of localization, but with high uncertainty in position. We define the initial detection as discovery. Reliable detection is required prior to and for the algorithms that enable a tracker or autonomous system to drive down position uncertainty. Human and automated detection methods are frustrated by low received power and uncertainties in radio pulse frequency, phase, and timing. In this presentation, we will review our work in that exploits the SDR continuous signal data record and adaptive near-optimal filtering to provide inputs to statistically robust decision criterion for pules detection. We will highlight the primary challenges in VHF signal detection and how our methods overcome these obstacles in a computationally efficient manner. Initial results will be presented that show pulse detections below the level of human hearing.
Using Unmarked Methods and Camera Traps for Estimating Abundance: A Guide to Study Design
M. Colter Chitwood; Gretchen H. Roffler; Anna K. Moeller; Paul M. Lukacs
Estimating abundance is a common need in wildlife management, and a plethora of methods exist to derive abundance. Unfortunately, many methods require expensive radiotags or dangerous fieldwork (e.g., helicopter flights) that often violate model assumptions anyway. Recently, advances in camera trapping technology and modeling approaches have facilitated the creation of unmarked methods for estimating abundance using camera trap images. Unfortunately, when it comes to study design and field implementation, these new methods potentially suffer from the same old problems, including misunderstood or ignored model assumptions that can result in spurious abundance estimates. Herein, we synthesize the most common unmarked abundance methods and provide guidance on how agencies can put various methods into practice. Using gray wolves (Canis lupus) on Prince of Wales Island, Alaska, USA, as a case study, we explore the costs and benefits of current unmarked, camera-based methods, including spatial capture-recapture, distance sampling, random encounter modeling, space-to-event, time-to-event, and instantaneous sampling. We discuss how and why managers should consider assumptions of each method, such that field implementation can provide quality data and trustworthy estimates. Thus, our synthesis serves as a decision-making tool for agencies who need to know abundance but need to choose an estimator with model assumptions, costs, and logistical field implementation in mind. At a time when camera trapping use is on the rise and modeling advancements are occurring rapidly, our synthesis serves as an important resource for conservationists to estimate and monitor abundance of populations, from small populations of conservation concern to large populations driven by fluctuations in hunter harvest.
Community Dynamics and Disturbance Responses of Tropical Arboreal and Terrestrial Mesomammals
Juliana Masseloux; Quy Tan Le; Jessica Burr; Brian D. Gerber
Tropical Southeast Asia is a biodiversity hotspot containing the highest proportion of endemic mammal species in the world. It also has the highest rate of forest loss from anthropogenic deforestation relative to any other major tropical region, experiences significant levels of illegal wildlife harvest, and has the greatest concentration of threatened mammal species. Where larger and more disturbance-sensitive species have been extirpated, frugivorous carnivores and arboreal mammals can play an important role in seed dispersal and ecosystem stability. However, there is a paucity of research on the ecology of small carnivores and non-primate arboreal mammals in tropical areas as they tend to be largely rare, cryptic, and nocturnal and thus difficult to survey. We conducted a multi-species multi-season study on spatio-temporal responses to anthropogenic disturbance by terrestrial and arboreal mesomammals – including the critically endangered Sunda Pangolin (Manis javanica). We set 75 terrestrial and 20 arboreal camera traps in Cat Tien National Park, Vietnam across a gradient of anthropogenic activities (e.g., distance to village, frequency of human use) and environmental characteristics (e.g., landcover type, forest structure). We identified 27 mammal species, including Sunda pangolin and 9 small carnivores (Family Carnivora). Several species were detected exclusively by arboreal or terrestrial cameras. Community occupancy of mesomammals was influenced by forest structure (basal area) and lower human-use frequencies, though individual species responses varied. Overall mammal richness for the protected area has also declined since previous biodiversity surveys were conducted in the late 1990s, highlighting ongoing threats to remaining wildlife populations. Our results contribute to the understanding of mesomammal responses to anthropogenic disturbance in Southeast Asia and identify conservation priority areas in Cat Tien National Park to curb anthropogenic disturbance levels and bolster populations.
Untrapped Potential: Do Bear Hunter Cameras Accurately Index Non-Target Species at Bait?
Ellen M. Candler; William J. Severud; Dean E. Beyer Jr.; Brian Frawley; Joseph K. Bump
Harvest reports are a commonly used index to monitor game species fluctuations. Hunter adoption of remote cameras opens possibilities for managers to collect citizen-derived data on multiple species using hunter surveys, but the effectiveness of this approach is largely untested. To examine whether hunter observations using remote cameras at black bear (Ursus americanus) bait sites and reported via hunter surveys are an effective method to record non-target species at bear bait sites, we compared experimental data to hunter observations from the same study area in the Upper Peninsula (UP), Michigan. We also quantified observations reported on hunter surveys as an index of density of gray wolves (Canis lupus), white-tailed deer (Odocoileus virginianus) and mustelid species (Mustelidae). We used Fisher’s exact test (α = 0.05) to test for differences in proportion of species observed and reported by hunters versus species observed at experimental bait for 2016-2018. We compared the camera index to wolf survey, white-tailed deer hunter, and mustelid trapper indices using Pearson’s correlation coefficients and fit linear models to each comparative relationship to elucidate the variation in density among counties that can be explained by the camera index. We detected no difference in hunter reported observations and experimental observations for 4 species, but hunters were over nine times more likely to report photographing wolves and nearly one third as likely to report photographing mustelids. We found significant correlations between camera index and deer hunter and mustelid trapper indices, but not for the wolf survey. Additionally, linear models accounted for variation in mustelid density, but not for deer or wolves. The widespread use of remote cameras by hunters, low-cost opportunity to expand hunter surveys, and potential to collect accurate community composition and relative indices data is strong evidence for the value of adding questions to hunter surveys regarding multiple species.
Can We Estimate the Abundance of Small Mammals with Camera Traps?
Marie I. Tosa; Damon B. Lesmeister; Matthew J. Weldy; Clinton W. Epps; Dana J. Morin; Taal Levi
Small mammals play an important role as herbivores, seed and fungal spore dispersers, and prey species in terrestrial ecosystems. Fluctuations in small mammal populations can predict changes to other components of the terrestrial food web that can be indicative of ecosystem health. Currently, live-trapping and mark-recapture methods are most commonly used to estimate small mammal abundances. These methods, however, typically only allow for one capture per trap, can cause physiological stress or death to captured animals, are labor and time intensive, and are costly, which restrict our ability to apply these methods over large temporal or spatial scales. We explored the possibility of using camera traps to estimate small mammal abundances across 8 small mammal grids that have been trapped and marked for long-term monitoring at the H. J. Andrews Experimental Forest in the Oregon Cascades during the fall of 2017. Here we analyzed data for Humboldt’s flying squirrel (Glaucomys oregonensis), Townsend’s chipmunk (Tamias townsendii), and deer mouse (Peromyscus maniculatus). We evaluated Spatial Count models, n-mixture models, and simple indices of relative abundance using camera trap data and compared these metrics to those derived from traditional mark-recapture data (i.e., Huggins robust design models, Spatial Capture Recapture models, and Spatial Count models). We also evaluated whether camera trap data can be calibrated using mark-recapture data. Results indicate that spatial count models alone are not sufficient to estimate small mammal abundances and that simple indices of relative abundance are just as informative as estimates from complex models. Camera traps, however, can detect a variety of species that cannot be detected using traditional mark-recapture methods and may be used to augment existing methods for quantifying the small mammal community.
Use of Aerial Surveys to Monitor Muskrat Population Trends in Northern Ohio
Catherine C. Dennison; Michael Ervin; Joe Barber
Many states, including Ohio, have historically relied on harvest data to assess trends in muskrat (Ondatra zibethicus) populations. However, fluctuations in fur prices and trapper participation can complicate these trends. In response to declining muskrat harvest and reports of declining populations of muskrats regionally and in Ohio, in the winter of 2013-2014 the Ohio Division of Wildlife initiated aerial muskrat house surveys on public wetlands in northern Ohio. These surveys provide a standardized method to track muskrat populations, independent of harvest. From the winter of 2013 to the winter of 2019 complete house counts were conducted on 118 survey units across 15 public land areas. Average muskrat house density across all units and years was 0.864 ± 0.091 houses/acre. House density was higher in coastal wetlands along Lake Erie (1.365 ± 0.144 houses/acre) then interior wetlands (0.244 ± 0.049 houses/acre) but varied considerably from unit to unit. Over the 7 years of surveys, average house density ranged from a high of 1.311 ± 0.329 houses/acre in the winter of 2015-2016 to a low of 0.501 ± 0.177 houses/acre in 2017-2018. Annual fluctuations on individual units were observed in response to management induced changes in vegetation availability. Our estimates of muskrat house density along Lake Erie are comparable to historic estimates in the region and indicate that muskrat populations on large wetland complexes in northern Ohio are holding steady. Maintaining and restoring quality wetland habitat is vital to maintaining healthy populations of this important wetland species.
Not All Ponds Are Created Equal: Determinants of Long-Term Lodge Occupancy Bybeavers
Glynnis A. Hood
Modeling habitat selection often represents single spatial scales and short time periods. However, for species that dramatically modify environments upon initial colonization, non-linear interactions change original habitat to one where habitat selection is confounded by subsequent modifications by the organism itself. The ability of beavers (Castor canadensis Kuhl, 1820) to dramatically alter habitat components over time makes them a model species for examining habitat selection spatially and temporally. This study identified habitat features selected by a relatively long-lived, colonial species, and how selection influenced site occupancy over varying spatial scales. Remote sensing and field data provided spatial and temporal coverage of four measures of occupancy (% occupancy over time, consecutive years occupied, consecutive years unoccupied, and number of lodges per pond). Models of occupancy over an 11-year period, combined with environmental variables, identified features influencing occupancy, abandonment, and recolonization at 30, 50, and 100-m scales. Unlike many studies, % deciduous was a poor predictor of occupancy. Instead, pond area, % grassland, distance to a pond that was active at least once, and standard deviation (SD) of slope were important variables. SD of slope was the only variable not influenced by beavers. Examining species-habitat relationships over broader spatial and temporal scales, particularly for ecosystem engineers, enhances understanding of non-linear effects and how they influence perceptions of habitat selection.
A Tale of Two Beavers: Are Translocated Beavers as Effective as Resident Beavers in Desert River Restoration?
Emma Doden; Phaedra Budy; Julie K. Young
Ecosystem engineers often drive increased species richness and greater habitat heterogeneity and can serve as a cost-effective, natural alternative to human restoration of degraded environments. American beavers (Castor canadensis) are considered ecosystem engineers because of their dam-building behavior, and as such they are being introduced to areas to restore degraded riparian habitats. Despite the popularity of beaver-assisted restoration, outcomes are variable and standardized best practices for translocation are lacking, especially in arid systems. In our study, we evaluate the efficacy of translocating beavers to desert restoration sites on the San Rafael and Price Rivers, Utah, by comparing their space use, survival rates, and activity patterns (e.g., dam building) to those of naturally occurring beavers. We hypothesized translocated beavers would initially use more space, have higher mortality, and build fewer dams than resident beavers, but that they would have similar patterns as resident beavers once territories were established. At their last points of detection, translocated beavers (n=11) had an average net squared displacement (NSD) of 26.9 km2 from their release location. Naturally occurring dispersing subadults (n=2) had a similar average NSD of 10.3 km2 from their release site, whereas adult resident beavers (n=4) had an average NSD of 0.15 km2 from their release location at their points of last detection. Predation-related mortality events occurred 19.7% more often for translocated beavers than naturally occurring beavers. No translocated beavers survived or remained detectable in the study area long enough to observe territory establishment or dam construction. To obtain establishment goals in these degraded arid streams, more beavers should be translocated to the sites coincident with the addition of structural features to encourage dam building (e.g., beaver dam analogs), and restoration areas should be expanded to account for large movements of translocated and dispersing beavers.

 

Virtual
Location: Virtual Date: Time: -