GIS, Telemetry, Remote Sensing II

Contributed Oral

Engineering for Wildlife: Collaborative Generation of Improved Monitoring Platforms
Staci Amburgey, Aadithya Prakash, Amy Yackel Adams, Shane Siers, Sarah Converse

Effective wildlife research and management are linked to the quality and quantity of data collected, but traditional data collection methods can be economically and logistically costly or infeasible. While we often leverage the use of cloud technology and open-source software to store and process data, we less often leverage advances in electronic hardware to solve our data collection challenges. This represents a lost opportunity. New sensor and embedded processor technologies have the potential to collect data more efficiently and affordably. Taking advantage of this opportunity requires cross-disciplinary partnerships to create innovative solutions, yet we often limit ourselves to our existing research networks and accustomed or proprietary methods. In the case of invasive brown treesnakes on the island of Guam, novel solutions to seemingly intractable monitoring and management challenges are needed. Abundance estimation has proved challenging due to the treesnake’s cryptic and nocturnal behavior but is essential to research and management decision making. In this partnership, we developed RePTaR (Remote PIT Tag Reader), a radio-frequency identification (RFID) system, to help us obtain individual snake identities to estimate abundance in areas targeted for suppression and native vertebrate reintroduction. RePTaR is built on the Arduino open-source electronics prototyping platform, allowing researchers with minimal engineering experience to easily debug and maintain the system. This system is substantially more cost effective and flexible than RFID systems currently in use. We detail the development process from conception through testing and discuss its importance for invasive species management. While our collaborative development of RePTaR resulted in a product specifically suited to this study system, it provides a framework for others wishing to develop better data collection systems for their target species. We highlight the need for interdisciplinary partnerships and collaborative approaches as a way to generate creative wildlife research solutions to surmount research and management challenges.

Landscape Structure of Woody Communities for Ocelots in Southern Texas
Jason “Jay” Lombardi, Humberto Perotto-Baldivieso, Maksim Sergeyev, Amanda Veals, Landon Schofield, John Young, Michael Tewes

Few ecological studies have explored landscape suitability using the gradient concept of landscape structure for wildlife species. Advancements in remote sensing and landscape ecological concepts now allow the examination of the full range of conditions that may be encountered by a species. Identification of conditions influencing the landscape ecology and population dynamics of threatened and endangered species allows for the development of more robust recovery strategies. Our objectives were to (i) identify the range of cover metrics associated with woody vegetation used by endangered ocelots (Leopardus pardalis), and (ii) quantify the potential distribution of usable woody cover patches for ocelots across southern Texas. We used the theory of slack and the gradient concept of landscape ecology combined with GPS telemetry data from 10 adult ocelots to identify the range of the spatial structure of woody cover in southern Texas. We estimated the amount of usable woody cover types and determined the transferability of this approach across years. Highly usable landscape-level patch structure for ocelots occurs in 28.1% of woody cover present in southern Texas. Spatial distribution of highly usable woody cover is comprised of large woody patches, with low shape-index values (1.07–2.25), low patch density (27.2–72.5 patches/ 100 ha), and low edge density (0–191.5 m/ha). Large patches of usable woody cover exist in the western and northeastern counties in southern Texas. Our study demonstrates a new approach for measuring vegetation cover for the development of a regional landscape use model for endangered ocelot populations in southern Texas. This range of landscape metrics identified suggests that there are more large patches of woody cover that contain the spatial structure used by ocelots than previously suspected. These results which are transferable across time and space will aid in informing future translocation and road mitigation efforts.

A New Citizen Science, Low-Cost Method to Map Degraded Habitat Across Sonoran Desert Landscapes for Restoration Planning
Helen Rowe, Dan Gruber, Mary Fastiggi

Restoration of degraded lands is important for protecting wildlife habitat and sustaining biodiversity. However, many managers lack the time and resources needed to systematically and comprehensively identify disturbed habitat to prioritize restoration efforts. We developed a novel, inexpensive, low-tech approach for training and engaging citizen scientists to identify degraded areas and recreational impacts within a defined unforested area. This citizen science restoration scanning process (hereafter CSRScan) follows four phases: 1) Landscape scans by citizen scientists using Google Earth Pro imagery; 2) A second scan of all marked sites using high resolution aerial photography; 3) Compilation of basic information about the degraded sites; 4) Addition of associated soil type and plant communities. In the 12,375 ha McDowell Sonoran Preserve (Scottsdale, Arizona) we detected 67 new sites not previously identified by land managers, using an estimated 235 citizen scientist hours and only 20 staff hours. Our process attributes important management considerations for each site, including distance from nearest access point, cause of degradation, and plant community and soil details. We verified accuracy of this method through independent field visits at 33% of the detected sites and confirmed details in all cases. We found that the CSRScan approach provided better perspective to accurately measure the scale and original source of degradation compared with field visits. The remote sensing approach can be conducted over a relatively short period of time using multiple citizen scientists; allows managers to undertake landscape level restoration prioritization and planning; and, if repeated, can be used to monitor changes in degradation and restoration over time as well as success of habitat restoration efforts.

Sources of Uncertainty in Interpretations of Seabird Species and Behavior from Aerial Imagery of Coastal Nesting Islands
Meredith Lewis, Logan R. Kline, Lauren Maher, Daniel Hayes, Cynthia Loftin

Long-term monitoring is integral to understanding avian population shifts and trends and informing species and ecosystem management practices. However, financial costs combined with relative remoteness, number of colonies, and disturbance serve as barriers to traditional survey approaches. Aerial surveys are an important tool for surveying colonial seabird and wading bird species, as large areas can be covered in a short period of time while minimizing disturbance, and imagery collected via planes provides a permanent record of survey. Even with high resolution imagery, however, observers can differ in their interpretation, misclassifying or completely missing subjects. When persistent, these errors can introduce bias into parameter estimates. We used a multiple-observer approach to count Herring Gulls (Larus argentatus), Great Black-backed Gulls (Larus marinus), and Double-crested Cormorants (Phalacrocorax auritus) in aerial imagery of 274 of Maine’s coastal islands. We aggregated observer interpretations of bird species and behaviors and assigned subjects to a class indicating complete agreement between observers, misclassification by one observer, or failed detection. We used multinomial regression analysis to determine factors that affected the probability of observer agreement and disagreement, including species, behavior, and habitat context. Image interpretation varied across observers, with > 25% of islands requiring additional interpretation to reconcile missed or incorrectly classified birds that affected total counts. While there often was agreement regarding species classifications between observers, behavioral state misclassification was a common error. For example, birds located in vegetated contexts were more likely to be misclassified behaviorally. Our work demonstrates the importance of accounting for errors in aerial image interpretation, as well as identifies characteristics of imagery that increase interpretation errors of coastal island nesting seabirds.

Some Useful Results for Applying Resource Selection Functions
Christen Fleming, Björn Reineking, Pati Medici, William Fagan, Justin Calabrese

Here we provide several practical results for the calculation of resource selection functions (RSFs), some of which also apply to species distribution models (SDMs) and step-selection functions (SSFs). First, we derive numerical error estimates for log-likelihoods that are approximated by the random sampling of “available” locations. This provides researchers with objective stopping criterion for convergent RSF estimates, with respect to the number of random locations sampled. Unfortunately, we find that for n observed locations, accurate RSF, SDM, SSF estimation requires O(n2) random samples, categorizing this as a computationally slow method. Second, we show how importance sampling can offer a large speed up in computation. Last, we provide a weighted log-likelihood approximation to adjust for temporal autocorrelation in RSF analysis. We demonstrate each finding on lowland tapir (Tapirus terrestris).

Cats and Dogs: Does Perceptibility Influence the Navigation of Risk and Reward?
Mitchell Brunet, Kevin Monteith, Katey Huggler, Justin Clapp, Daniel Thompson, Patrick Burke, Mark Zornes, Patrick Lionberger, Miguel Valdez, Joseph Holbrook

Animals may directly perceive the presence of risk via the visualization of a predator, or may indirectly perceive risk via the visualization of habitats used by a predator. We investigated the response of a mesopredator to the presence of risk and reward imposed by an apex predator, where risk and reward vary in relative perceptibility, and assessed whether direct or indirect assessments of risk and reward better explain the navigation of those sources. First, we evaluated the response of coyotes (Canis latrans) to direct encounters with mountain lions (Puma concolor), where coyotes compose 8.4% of prey killed by mountain lions. Second, we assessed the use of carrion at mountain lion kill sites by coyotes. Third, we tested whether direct or indirect assessments of mountain lions and their kill sites better explained the way coyotes navigate. Coyotes did not respond to the risk of direct encounters with mountain lions. Instead, coyotes facilitated encounters with mountain lions through their movement. Coyote use at mountain lion kill sites increased initially with time from the kill, and coyotes navigated risk and reward based on indirect assessments. Despite direct assessments more accurately characterizing the arrangement of risk and reward, their use for behavioral assessment in our study was limited, because indirect inferences more broadly characterized the arrangement of risk and reward across the landscape.

Contributed Oral
Location: Virtual Date: November 4, 2021 Time: 2:00 pm - 3:00 pm