<<

THE WILDLIFE SOCIETY Remotely Wild

Newsletter of the Spatial Ecology & Telemetry Working Group

Issue 33 Spring 2020

From the Working Group Chair A Difficult Time:

We bring you this newsletter at a difficult time. As the world responds to the Coronavirus pandemic, we all go through our own shocks as we adjust and prepare. We worry about our health, and even more for our loved ones. We worry about the economy and our future, as businesses shut down and our 401Ks tumble. We worry about our jobs and our work, and our ability to pay the bills.

In wildlife management and research, we are fortunate in that we often work in more isolated conditions than most professions. We are unlikely to contract this disease when we sit alone on a mountain, watching in fascination at a world that amazes us. But many of us work in jobs that have regular contact with the public, or are in school or teach. We face the same threats the rest of the world does.

As schools shut down or move online, we who are parents are adapting to life with children who are now home all day. Those of us who teach struggle to shift our classes online. Those of us who are students worry about our education. Conferences and professional meetings are of course shutting down, and rightly so. Mariah Simmons, TWS Unit Services Manager, is maintaining a running list of TWS event cancellations and postponements that you can check at the link below. Note that there are separate tabs describing regional travel restrictions and events impacted.

Social Isolation:

As you all know, governments around the world are taking aggressive steps to contain this virus. We are all learning about “social distancing.” This pandemic will pass, though, and the social isolation we are imposing on ourselves will reduce the damage and save lives. The downside of social isolation, of course, is that it isolates, and we are a social species. We may all be well-advised to reach out occasionally, to colleagues and loved ones, just to check in and reconnect. Please know that you are not alone as we collectively face this global threat.

What can we do to help?

Here at SETWG, we will try to keep our little part of the world going. In this newsletter you will find some interesting articles from our 2019 travel grant recipients, describing their research in movement ecology, carnivore interactions, and effects of sampling methods on defining population sizes. We also have some announcements about our SETWG awards program, the 2020 national TWS conference in Kentucky, and upcoming elections. We will also be in touch more regularly with you in the future. Do you have ideas that could help our members? If so, please let us know! We will be happy to share with the working group. For example, GIS professors struggling to transition to online courses may be interested this resource from the Science Education Resource Center (SERT) and the National Association of Geoscience Teachers (NAGT): https://serc.carleton.edu/teachearth/ teach_geo_online/index.html

Thank you for reading, and thank you for being a part of the Spatial Ecology and Telemetry Working Group! Best regards, Jeff Jenness ([email protected]) TWS Event Cancellaons & Postponements: hps://docs.google.com/spreadsheets/u/2/d/e/2PACX-1vQoHjh- rH5TKdJaJz4sXnuX-wMCli8OOIXf0nxf9wzO6fzI-llqcE5MwwDzu-UFdknn71G2t7eCZdA-/pubhtml?urp=gmail_link SETWG

In this issue: Fg 1: From the Working Group Chair. Pg 2: SETWG annual Treasurer’s report Pg 3: Conference announcement & executive team update Pg 4: SETWG Awards - solicitation for nominations Pg 5: Movement Ecology tools Pg 6: How you define your population matters: Making inference about habitat associations varies depending on the sampling method used Pg 11: Modeling Carnivore Interactions in Northeastern Oregon Using Activity & Occupancy Patterns Pg 17: Movement ecology and space-use of mountain lions in wild

Treasurer’s Report at a glance: 2020

Balance as of 1/1/2019 $5,469 2019 Total Income $3463 2019 Total Expenses $3100

Current balance $5,832

Membership 224

WHERE IN THE WORLD ARE SETWG MEMBERS..?

Remotely Wild - Spring 2020 2 SEQUOIA CLUB

2020 National TWS Conference The 2020 National TWS conference is currently scheduled for September 27 – October 1, 2020, in Louisville, Kentucky. Naturally this may change depending on the status of the coronavirus pandemic, but for the time being we will hope for the best and assume it will proceed as planned. Travel Grants:

SETWG traditionally provides a few travel grants to students and early professionals to help them make the trip, and to help them participate in our professional community. We will make an announcement with the details soon, but for the time being please let your colleagues and/or students know this is in the works.

Conference Events: We have two events proposed for the conference, and which are currently being decided on by the council. If approved, we will be sponsoring a pair of workshops teaching us how to use the Google Earth Engine, and a special poster session illustrating innovative uses of GIS in wildlife management and research. We’ll be in touch with details when we hear back from the council.

We are also supporting a symposium on long-term datasets proposed by the Biological Diversity Working Group. This, too, will be decided by the TWS council, but we have our fingers crossed for BDWG. 2020 SETWG Elections The current set of working group officers are nearing the end of their 2-year terms and it is time to pass the torch. Would you like to serve as a working group officer? We have a large and dynamic working group and this is a great opportunity to get involved. The working group needs you and you’ll enjoy the work! The Spatial Ecology and Telemetry working group has three officers. All positions are 2-year terms and all are up for election. If you would like to know more about the duties involved in a particular office, please contact the current officer and we’ll be happy to discuss it:

Chair: Jeff Jenness, [email protected] Treasurer: Kathy Zeller, [email protected] Secretary: Alex Wolf, [email protected] To be a candidate, please send your name, the office you are running for, a brief biography, and a sentence or two explaining why you want to run, to Alex Wolf at [email protected]. The deadline for submissions is August 28, 2020. We will compile all the candidate information and send it to the working group, and then hold online elections for a couple of weeks and announce the new officers before the national TWS conference in Louisville.

Remotely Wild - Spring 2020 3 SETWG SETWG Awards 2020 Solicitation for nominations:

The Spaal Ecology and Telemetry Working Group is excited to announce that we are accepng nominaons for 2020 awards that recognize professionals in the field of GIS, Spaal Ecology, Remote Sensing & Biotelemetry who have made significant contribuons to the field of wildlife biology. Award recipients do not need to be wildlife biologists or even involved in any environmental research or management. They only need to have wrien or produced something, or provided some service that has substanally improved our ability to do our job and enabled us to do things we may not have been able to do before. Past recipients include the developers of soware packages such as the adeHabitat and unmarked packages for R, the Home Range Tools and ArcMET extensions for ArcGIS, Circuitscape, and Fragstats, among many others.

Although our awards do not include any kind of cash prize, they are a way for us, as a professional society, to say thank you to these individuals for the help they have given us. If you believe that a person or a group deserves special recognion for a service that they have provided the spaal ecology community, please send an email with their name, affiliaon and a brief descripon of why they should receive an award to the 2020 SETWG award commiee: [email protected]

4 SEQUOIA CLUB Movement Ecology Highlighting R-based tools for analyses

Movement ecology is a quickly growing subdiscipline that combines both of our favorite things here at SETWG, spaal ecology and telemetry. There is now a dedicated journal, Movement Ecology, for research specifically focused on this topic. There was an enre Gordon Research Conference on movement ecology this year. There have also been special issues in journals and symposia at society meengs dedicated to this topic. And there was much interest in the habitat selecon and movement ecology workshop we sponsored at this year’s annual TWS conference (and a request from members for yet another one next year).

The field is growing in part due to the plethora of telemetry data that is being collected (and made available through sites like MoveBank), and in part due to our increased capacity to analyze these data in complex and interesng ways. This burgeoning of techniques for analyzing movement data can be overwhelming as can the sheer number of R packages available to employ these techniques. Because of this, we wanted to highlight a recently published paper in The Journal of Animal Ecology entled Navigang through the R packages for movement.

Rocío Joo and co-authors reviewed 58 R packages that can be used to analyze movement data. They cover the gamut of opons for movement data analysis, outline the packages that can be used for each one, and even provide a handy reference table. For example, they cover packages that can be used in data pre-processing for data like global locaon sensor data, radio-tags, and accelerometry and magnetometry data. They idenfy packages that offer funcons for data cleaning, data exploraon, and calculang metrics like step lengths and turning angles. They write about categorizing movement data into units like steps or paths, or categorizing the data into different behavioral states with hidden Markov models. Then, they go on to describe packages that can be used in analyses like home range esmaon, habitat use, and interacons among tagged individuals. They also provide informaon on packages that simulate movement trajectories and allow for visualizaon, like creang animaons of animal movement from collar or other tag data. William C. Campbell They helpfully point out package dependencies, for example, objects or funcons from one package are needed as inputs to another package.

If you are just geng started with movement ecology or are seasoned at it, this paper is definitely worth a read and helps sort out the somemes confusing array of R packages that are available to analyze tracking data in a handy reference table.

Remotely Wild - Spring 2020 5 SETWG How you define your population matters: Making inference about habitat associations varies depending on the sampling method used Sarah B. Bassing, PhD student University of Washington [email protected]

I received a student travel grant from the Spaal Ecology and Telemetry Working Group to aend the American Fisheries Society and The Wildlife Society 2019 Joint Annual Conference in Reno, NV this fall. I was honored to receive this award and thrilled to give an oral presentaon entled “Potenal biases in camera trap data when monitoring mulple species," on behalf of my coauthors at the conference. This research is part of the Washington Predator-Prey Project, a collaborave project between the University of Washington and the Washington Department of Fish and Wildlife. More about the larger study can be found at hps://predatorpreyproject.weebly.com/

Remotely Wild - Spring 2020 6 SETWG

Camera traps are widely used for monitoring wildlife species. Although camera data can yield inferences about resource selecon and other metrics of interest similar to those made from other data collecon methods (e.g., telemetry), most comparave studies focus on a single species specifically targeted by the camera trapping. Camera traps are increasingly being used for mul-species monitoring but camera placement and sampling design can strongly influence the frequency with which different species are detected within a community. We were therefore interested in evaluang whether camera traps provide data that are equally representave of habitat use for mulple species when used as a monitoring tool for both predators and prey.

Our queson was inspired by the fact that we, like many camera trap-users, placed cameras in a nearly random array in an effort to collect occurrence data on a community of species differing in body size, foraging behavior, and guild. I say nearly random because we used a strafied random sampling design to place camera traps across two study areas in eastern Washington, USA. But at the microsite-level, we placed each camera trap on a primive road or game trail to increase the likelihood of detecng elusive species (i.e., predators) instead of in truly random locaons. Given that predator and prey species may use these travel routes and the habitats they bisect differently, we hypothesized that the departure from true random placement may bias detecon and inferred habitat use for some species of interest. Specifically, we predicted that camera trap data would misrepresent habitat associaons for herbivores (i.e., , Odocoileus hemoinus) more so than for carnivores (i.e., , Puma concolor) since our placement of cameras was more targeted at the microsite-level for carnivores.

7 SETWG

We used photo-capture data collected by 120 camera traps from June 1 – Sept. 30, 2018 to test our hypothesis. We strafied our camera placement by elevaon but capped our highest strata at 2,100 meters to avoid sampling alpine habitat primarily for logiscal reasons. We populated single-species, single-season occupancy models with detecon data for cougars and mule deer, respecvely, and fit the species-specific models using a number of covariates we hypothesized would be predicve of habitat use (e.g., land cover type, elevaon, terrain ruggedness, road density, stream density, etc.). We paired this analysis with resource selecon funcon (RSF) analyses using GPS locaon data generated during summer 2018 from 22 and 80 mule deer collars, respecvely.

The GPS collars were deployed within the same study areas as the camera traps with the intent of sampling the same populaons using mulple monitoring methods. We included the same covariates in the RSFs as we did the occupancy models, and added a random effect for individual. We ran models for each study area separately and used AIC for model selecon with both the occupancy models and RSFs. Finally, we compared results from the camera trap-based occupancy models to the GPS collar-based RSF models for each species. We expected greater discrepancies in the results from the two methods for a given species if camera trap placement biased the detecon data, and thus esmated habitat associates, of that species.

We found elevaon was an important covariate for both species in both study areas, regardless of the data source and analycal method used. Interesngly, we found the relaonship between elevaon and habitat use/selecon differed between methods for both species, most notably with cougars. In the West study area, the camera trap- based occupancy model indicated a quadrac effect of elevaon on cougar habitat use, where probability of use declined at higher elevaons, whereas the GPS collar-based RSF indicated cougar habitat selecon increased exponenally with elevaon. This relaonship flipped in the East study area, where the probability of habitat use for cougars increased with elevaon but relave probability of selecon decreased with elevaon. As for mule deer in the West study area, we found habitat use and selecon were both posively associated with elevaon but that the relaonship was linear with the camera trap data and quadrac with the GPS collar data.

Remotely Wild - Spring 2020 8 SETWG

Although there appears to be a larger discrepancy between the occupancy model and RSF results for cougars compared to those of mule deer, we are hesitant to interpret these results as a species-specific bias in the camera trap data. These discrepancies may be due to a number of sampling issues that have nothing to do with our original concern that camera placement on roads and game trails may bias detecon of some species. Most notably, the GPS collar data indicated that both cougars and mule deer were selecng for high elevaon habitat in the West study area, well about the 2,100 m cutoff we used for placing camera traps. In addion, many of the collared animals where restricted to small porons of our study areas or moved outside our pre-defined study area which le large areas unsampled by GPS collars. As a result, we effecvely sampled different populaons depending on the sampling method used which may explain the discrepancies we observed. Essenally, the camera traps allowed us to sample cougars and mule deer at specific locaons and relate those observaons to characteriscs of the study areas whereas the GPS-collars allowed us to measure characteriscs associated with the captured cougar and mule deer populaons.

Although we were unable to fully answer our original research queson with these data, our results do highlight the fact that how you define and sample your populaon greatly influences the inference you make. We hope to evaluate these sampling issues further as we connue to collect more camera trap and GPS collar data over the coming years. We plan to expand this analysis to include addional species (e.g., wolves, white- tailed deer, ) and consider how seasonality influences the potenal differences in sampling methods.

9 SETWG

Remotely Wild - Spring 2020 10 SETWG Modeling Carnivore Interactions in Northeastern Oregon Using Activity & Occupancy Patterns

Rylee Jensen1, Joel S. Ruprecht2, Tavis D. Forrester3, Elizabeth K. Orning2, Taal Levi2, Darren E. Clark3

1Utah State University, Department of Wildland Resources 2Oregon State University, Department of Fisheries & Wildlife 3Oregon Department of Fish & Wildlife

In mul-carnivore systems, carnivores must forage and reproduce while compeng with other carnivores (Linnell and Strand 2000, Hearn et al. 2018) and possibly avoiding intraguild predaon (Polis et al. 1989, Palomares and Caro 1999). Most research has been focused on pairs of compeng species and it is unclear how carnivores interact with mulple competors, especially in systems with subordinate and apex predators. In this study we assessed the degree of co- occurrence between cougars (Puma concolor), bobcats (Lynx rufus), and coyotes (Canis latrans) in northeast Oregon using occupancy models based on species-specific detecon rates. In addion, we also accounted for habitat covariates (e.g. terrain ruggedness, canopy cover, distance to water sources, etc.) that may signify preferences for certain areas. Our objecve was to first explore how these environmental covariates affect the occupancy of these three species, then examine how this influences pairwise species interacons (i.e. whether occupancy changes as a funcon of the presence of competors). We also modeled the influence of anthropogenic detecons on species interacons. Studying interacons between these sympatric carnivores as well as their spaotemporal and environmental relaonships may elucidate to what degree these interacons occur and what factors may contribute to niche paroning or compeon.

Remotely Wild - Spring 2020 11 SETWG

Our two study sites were located in the Blue Mountains of northeastern Oregon. The first, the Starkey Experimental Forest, is a 25,000-acre ungulate-proof enclosure located approximately 30 miles southwest of La Grande, Oregon (Figure 1). It was established in 1989 by the U.S. Forest Service to research the long-term effects of land management pracces on domesc cale (Bos tarus) and nave ungulates like elk (Cervus elaphus) and mule deer (Odocoileus hemionus). The 8-. tall perimeter fence does not inhibit carnivores’ movements in and out of the enclosure, so Starkey connues to funcon as a natural system despite this. Other carnivore species present here apart from our three main study species include black bears (Ursus americanus), mustelids such as striped skunks (Mephis mephis), American badgers (Taxidea taxus), and weasels (Mustela spp.), and occasionally red foxes (Vulpes vulpes) and gray wolves (Canis lupus). Common prey species of these carnivores include tree squirrels (Tamiasciurus spp.), woodrats (Neotoma spp.), and snowshoe hares (Lepus americanus).

Fig. 1: . Map and scale of the Starkey Experimental Fig. 2: Map and scale of the Mt. Emily Wildlife Management Forest in northeastern Oregon (from Kie 2013). Unit in northeastern Oregon (from Davidson et al. 2014). The grid displayed from this 2014 study is not the same in our study.

The other study area is the Mt. Emily Wildlife Management Unit (Figure 2), located approximately 12 miles northeast of Starkey. This area covers 770 square miles and has similar wildlife assemblages and dominant habitat features to Starkey. However, one important disncon between this study area and Starkey is the presence of the Mt. Emily wolf pack, whose territory encompasses our camera trap grid within this area. Wolves were occasionally sighted in Starkey, but no pack was established within the perimeter fence.

12 SETWG

94 camera traps were placed on a grid with 1 km. spacing in the Starkey Experimental Forest from May-September in 2016 and April-September in 2017. 54 cameras were placed in Mt. Emily from August-November in 2016 and 2017; due to this study area being twice as large as Starkey, cameras on this grid were placed 2 km. apart. Cameras were unbaited and placed approximately 1 . off the ground in Starkey and slightly higher in Mt. Emily. They were commonly deployed on game trails, drainages, stream beds, and ridges to opmize detecon of carnivore species. Photos were then tagged in DigiKam to determine baseline detecon rates for our target species.

Habitat covariates consisted of detecon (observaon/occasion) covariates and occupancy (site) covariates. Detecon covariates are those that influence whether a species is able to be detected by our camera traps at a parcular site (given that they are occupying the site), whereas occupancy covariates are those that influence a species’ use of a site (MacKenzie et al. 2002). We determined seven detecon covariates (road, trail, other, percentage canopy cover, Julian day of deployment, project site, and camera elevaon) and eleven occupancy covariates (year, vegetaon greenness, terrain ruggedness, distance to nearest forest edge, number of wolf detecons, distance to nearest stream, distance to nearest open road, distance to nearest perennial spring, number of springs within 1 km radius, number of human detecons, and number of vehicle detecons).

Covariate layers were generated in ArcMap 10.6.1. All covariates were standardized for analysis and tested for covariance; any two covariates with a covariance >0.5 were omied from the same model together. For analysis purposes, we omied data from Starkey from April-June to reduce detecon bias with potenally overlapping the denning season for coyotes. We used 14 survey occasions of 7 days, which began on July 1 for Starkey and August 5 for Mt. Emily.

Binary encounter histories were generated for each of the sample periods. Mulspecies occupancy models were run in MARK (Rota et al. 2016) to test the effects of various combinaons of environmental covariates and species interacons. This generated a probability matrix of occupancy in the same sample periods given detecons. Top models were then selected based on AICc score and weight.

We had four main hypothesis scenarios concerning how habitat covariates and species interacons would impact occupancy: 1 — Species interacons alone explain the occupancy of all three species. 2 — Habitat preference supersedes species interacons in explaining species occupancy. 3 — Habitat preference has a large effect on occupancy, but this is also influenced by species interacons. 4 — Species occur independently (i.e. habitat preference and spp. interacons have no effect).

Remotely Wild - Spring 2020 13 SETWG We predict that coyotes and bobcats may exhibit some avoidance behavior of cougars in space and me due to the risk of intraguild predaon (Hass 2009, Koehler & Hornocker 1991). However, there is also the possibility that coyotes may be aracted to cougars to some degree for scavenging opportunies. Cougars and bobcats may exhibit lile interacon in addion because of differing habitat and dietary preferences (Hass 2009). Coyotes and bobcats, on the other hand, exhibit a fair amount of dietary overlap and have comparable home range sizes (Witmer and DeCalesta 1986, Litvais and Harrison 1988, Neale and Sacks 2001), suggesng niche paroning in areas they co-occur.

Addionally, we hypothesized three more scenarios concerning human influence on species interacons: I. Human and vehicle detecons have no effect on species interacons II. Human detecons have a greater effect on species interacons than vehicle detecons III. Vehicle detecons have a greater effect on species interacons than human detecons

We expect our results to appear similar to Rota et al.’s (2016) mulspecies occupancy models involving interacons between bobcats, coyotes, gray foxes (Urocyon cinereoargenteus), and red foxes (Vulpes vulpes). Separate figures were generated to demonstrate the influence of habitat covariates (Figure 3) and species interacons (Figure 4). This study found that the spaal distribuon of species was influenced by both factors; for example, coyotes and gray foxes were more likely to co-occur at sites with high human disturbance, but less likely at sites with less human disturbance (Rota et al. 2016). This is why it is crucial for us to include species interacons as a funcon of human- related detecons.

Fig. 3: . “Marginal occupancy probability of bobcat (Lynx rufus, row 1), coyote (Canis latrans, row 2), grey fox (Urocyon cinereoargenteus, row 3) and red fox (Vulpes vulpes, row 4) in the Mid-Atlanc USA as a funcon of the average number of hikers photographed per day (column 1), housing density within a 5 km radius (column 2) and the proporon of area recently disturbed within 5 km (column 3). Solid lines represents the mean posterior distribuons, and grey ribbons envelop 95% credible intervals. All variables not included in a plot are assumed fixed at their observed mean.” (from Rota et al. 2016) Remotely Wild - Spring 2020 14 SETWG We predict that coyotes and bobcats may exhibit some avoidance behavior of cougars in space and me due to the risk of intraguild predaon (Hass 2009, Koehler & Hornocker 1991). However, there is also the possibility that coyotes may be aracted to cougars to some degree for scavenging opportunies. Cougars and bobcats may exhibit lile interacon in addion because of differing habitat and dietary preferences (Hass 2009). Coyotes and bobcats, on the other hand, exhibit a fair amount of dietary overlap and have comparable home range sizes (Witmer and DeCalesta 1986, Litvais and Harrison 1988, Neale and Sacks 2001), suggesng niche paroning in areas they co-occur.

Addionally, we hypothesized three more scenarios concerning human influence on species interacons: I. Human and vehicle detecons have no effect on species interacons II. Human detecons have a greater effect on species interacons than vehicle detecons III. Vehicle detecons have a greater effect on species interacons than human detecons

We expect our results to appear similar to Rota et al.’s (2016) mulspecies occupancy models involving interacons between bobcats, coyotes, gray foxes (Urocyon cinereoargenteus), and red foxes (Vulpes vulpes). Separate figures were generated to demonstrate the influence of habitat covariates (Figure 3) and species interacons (Figure 4). This study found that the spaal distribuon of species was influenced by both factors; for example, coyotes and gray foxes were more likely to co-occur at sites with high human disturbance, but less likely at sites with less human disturbance (Rota et al. 2016). This is why it is crucial for us to include species interacons as a funcon of human- related detecons.

Our preliminary results indicate that models incorporang species interacons along with the top-ranking occupancy and detecon covariates for each of the three species individually explained species occupancy beer than models that did not incorporate species interacons. Our top detecon model was non-me varying (i.e. each species had an equal probability of detecon across all sample periods) and included road, trail, and canopy cover as the best combinaon of detecon covariates for all species.

Fig. 4: Occupancy probability of bobcat (Lynx rufus), coyote (Canis latrans), grey fox (Urocyon cinereoargenteus) and red fox (Vulpes vulpes) condional on the presence and absence of each of the other species. The occupancy probability of the species in each column is condional on the presence and absence of the species in each row; for example, the plot in column 1, row 2 represents bobcat occupancy probability condional on the presence and absence of coyote. Lines represent posterior means, and ribbons envelope 95% credible intervals. All variables not included in a plot are assumed fixed at their observed mean. Addionally, condional plots are marginalized over the 2 species that did not occur in a plot; for example, the plot in column 1, row 2 sums over all combinaons of grey fox and red fox presence and absence. (from Rota et al. 2016)

Remotely Wild - Spring 2020 15 SETWG Terrain ruggedness and count of wolf detecons explained both bobcat and coyote occupancy best (individually), while year and count of wolf detecons were ranked highest for cougars. However, a separate model excluding any covariates for cougars but keeping the top-ranking occupancy covariates for bobcats and coyotes also ranked extremely high, just <1 AICc difference from the previously stated model. This suggests that the top occupancy covariates for cougars may not be significant overall and it is likely that other factor(s) we did not model (e.g. mule deer distribuon/movement) may be driving their occupancy at each of our study sites. More work is also needed to accurately interpret how count of wolf detecons influences occupancy at camera sites due to significantly more detecons in the Mt. Emily study area than in Starkey.

Future work will also determine the effect of human and vehicle detecons on species interacons at camera sites. Although our study areas have minimal human disturbance aside from roads, carnivores may sll be aracted or repelled by anthropogenic features; for example, the presence of campers may indicate food sources that some carnivores (especially bold, resilient mesocarnivores like coyotes and bobcats) can take advantage of. Dirt roads in parcular may facilitate movement for our study species (e.g. cougars, Dickson et al. 2005) or present a barrier for movement (e.g. bobcats, Young et al. 2019). Addionally, each of our three study species can be hunted in northeastern Oregon several or all months of the year, which in turn could create avoidance paerns around heavily trafficked areas like open roads. Our results may aid managers in accurately determining where these species are

Literature Cited:

Davidson, G.A., Clark, D.A.C., Johnson, B.K., Waits, L.P., and J.R. Adams. 2014. Esmang cougar densies in northeast Oregon using conservaon detecon dogs. Journal of Wildlife Management 78(6): 1104-1114. Dickson, B.G., Jenness, J.S., and P. Beier. 2005. Influence of vegetaon, topography, and roads on cougar movement in southern California. Journal of Wildlife Management 69(1): 264-276. Hass, C.C. 2009. Compeon and coexistence in sympatric bobcats and pumas. Journal of Zoology 278(3): 174–180. Hearn, A. J., Cushman, S. A., Ross, J., Goossens, B., Luke, T., Hunter, B., and D.W. Macdonald. 2018. Spao-temporal ecology of sympatric felids on Borneo. Evidence for resource paroning? PLoS ONE 13(7), e0200828. Kie, J.G. 2013. A rule-based ad hoc method for selecng a bandwidth in kernel home-range analyses. Animal Biotelemetry 1:13. Koehler, G.M. and M.G. Hornocker. 1991. Seasonal Resource Use Among Mountain Lions, Bobcats, and Coyotes. Journal of Mammalogy 72(2): 391–396. Linnell, J.D.C. and O. Strand. 2000. Interference interacons, co-existence and conservaon of mammalian carnivores. Biodiversity Research 6: 169–176. Litvais, J.A. and D.J. Harrison. 1989. Bobcat–coyote niche relaonships during a period of coyote populaon increase. Canadian Journal of Zoology 67(5): 1180–1188. MacKenzie, D.I., Nichols, J.D., Lachman, G.B., Droege, S., Royle, A., and C.A. Langmm. 2002. Esmang site occupancy rates when detecon probabilies are less than one. Ecology 83: 2248-2255. Neale, J.C.C. and B.N. Sacks. 2001. Resource ulizaon and interspecific relaons of sympatric bobcats and coyotes. Oikos 94(2): 236–249. Palomares, F. and T.M. Caro. 1999. Interspecific Killing among Mammalian Carnivores. The American Naturalist 153(5), 492–508. Polis, G.A., Myers, C.A., and R.D. Holt. 1989. The Ecology and Evoluon of Intraguild Predaon: Potenal Competors That Eat Each Other. Annual Review of Ecological Systems 20: 297–330. Rota, C.T., Ferreira, M.A.R., Kays, R.W., Forrester, T.D., Kalies, E.L., McShea, W.J., Parsons, A.W., and J.J. Millspaugh. 2016. A mulspecies occupancy model for two or more interacng species. Methods in Ecology and Evoluon 7(10): 1164–1173. Witmer, G.W. and D.S. deCalesta. 1986. Resource use by unexploited sympatric bobcats and coyotes in Oregon. Canadian Journal of Zoology 64(10): 2333–2338. Young, J.K., Golla, J., Draper, J.P., Broman, D., Blankenship, T., and R. Heilbrun. 2019. Space Use and Movement of Urban Bobcats. Animals (Basel) 9(5): 275.

Remotely Wild - Spring 2020 16 SETWG Movement ecology and space-use of mountain lions in wild West Texas

Dana L. Karelus & Patricia Moody Harveson Borderlands Research Institute, Sul Ross State University, PO Box C-21, Alpine, TX, 79832, USA

Understanding the movements of large carnivores across the landscape and their corresponding space use provides us with important details for their management, the management of their prey, for land-use and corridor planning, and for studies on their populaon dynamics and densies. However, obtaining this informaon can be difficult because large carnivores oen require large areas, occur in low densies, and can be at odds with humans (Ripple et al. 2014; Pitman et al. 2015). Fragmented landscapes and predator control efforts may make movements within an animal’s home range and among subpopulaons more difficult, thereby potenally affecng their overall space-use and intensifying the risks associated with dispersal. Mountain lions (Puma concolor) in West Texas face this scenario.

Remotely Wild - Spring 2020 17 SETWG

West Texas is part of the ; land cover types range from desert scrub to grassland savannah to rugged, rocky terrain and to forests in the higher elevaon mountains. The in Big Bend Naonal Park (BIBE) along the U.S.-Mexico border, and the Davis Mountains, approximately 100 km northwest of BIBE, are two such mountain ranges in the southern poron of the region (Figure 1); termed sky islands because the forests in these mountains are fragmented by a sea of desert or grassland. The main subpopulaons of mountain lions in the southern poron of the region occur on these sky islands. Naonal Park lies approximately another 100 km north of the Davis Mountains and also supports a subpopulaon of mountain lions (Harveson et al. 1999), however a major east-west interstate, I-10, runs between the two ranges and likely acts as a barrier to animal movement. In the southern poron of West Texas (south of I-10; Presidio, Jeff Davis, and Brewster counes), the roads have far less traffic and the human populaon density is very low (esmated at 0.58 people/km2; United States Census Bureau 2018). As with all of Texas, almost all of the land in this region is privately owned and some of the largest ranches in Texas are located in the area. This means that, aside from I-10, roads tend to be a low cause of mortality for mountain lions here. However, humans are sll their largest cause of mortality. In Texas, mountain lions are considered a non- game species, meaning that they may be trapped or hunted at any me of the year without limit. Indeed, the greatest sources of mortality for mountain lions captured on state or federal properes in the greater region (Guadalupe Mountains Naonal Park, Carlsbad Caverns Naonal Park, BIBE, and Big Bend Ranch State Park) were lethal predator control and hunng, mainly when animals went outside the park boundaries onto private lands (Young et al. 2010; Harveson et al. 2012). The associated mountain lion survival rates were among the lowest reported in the U.S (Piman et al. 1999; Young et al. 2010; Harveson et al. 2012).

Within the state and naonal parks, mountain lions receive at least some protecons, but because parks are not large enough or in close enough proximity to support the enre mountain lion populaon of West Texas, whole subpopulaons must live enrely on private lands. These animals may face higher mortality risk than those on park properes, but also potenally may experience less disturbance from people (i.e. no park visitors) and the land and their prey species are managed differently. These factors not only have survival consequences, but also might affect mountain lion movements and space use, yet no such study had previously been performed on private lands in the region. Addionally, dispersal among these subpopulaons and others in Mexico and New Mexico is likely crical for their genec health (Holbrook et al. 2012). Therefore, we sought to invesgate these aspects of mountain lion ecology for those in the subpopulaon on the private lands in the Davis Mountains and the subpopulaon in Big Bend Naonal Park, as well as movements of dispersers throughout the region.

Fig. 1: Map of the study areas in the Davis Mountains and Big Bend Naonal Park in the southern poron of the West Texas region in the United States of America.

Remotely Wild - Spring 2020 18 SETWG

Mountain lion captures for this study occurred from 2011 through 2017, and resulted in the collaring and tracking of 27 subadult and adult individuals (2 females and 2 males in BIBE; 15 females and 8 males in the Davis Mountains). Through the years, various brands of GPS collars were used (Advanced Telemetry Systems G2110L and G2110E collars, Isan, MN; North Star NSG-D1 collar, King George, VA; Vectronic VERTEX Plus GLOBALSTAR, Berlin, Germany) and fix schedules were adjusted. In some cases, collars were set to collect locaons at more frequent intervals at certain mes. We chose to follow the framework by Calabrese et al. (2016) and used connuous me movement models and autocorrelated kernel density esmator (AKDE) to analyze and esmate home ranges (Fleming & Calabrese 2017; Noonan et al. 2019). In this way, we were able to account for the inherent autocorrelaon among an individual’s locaons and we avoided discarding data due to the variable fix schedules (Fleming et al. 2018). We used program R (R Core Team 2019) and the package ‘ctmm’ (Fleming & Calabrese 2019) to run our analyses.

Fig. 2: Examples of the path trajectory (top row) and variograms plot (boom row) from A) a range resident mountain lion (Puma conolor) and B) from a disperser. For resident animals, the semi-variance will reach a general asymptote whereas for dispersers, the semi-variance will connue to increase.

Remotely Wild - Spring 2020 19 SETWG

Based on variograms of the mountain lions’ movements, we defined 3 subadults as dispersers (Figure 2). These 3 individuals (2 females, 1 male) were all captured in the Davis Mountains and were on relavely long trips compared to assumed exploratory trips by other subadults in the study. One disperser and 1 subadult on a quick exploratory trip came within at least 2-3 km of I-10, but both turned back at that point. Aer approximately 1-2 months of leaving the Davis Mountains (straight-line distances from start to end ranging between approximately 40 km and 135 km), the 2 female dispersers were trapped and killed and the male’s movements stopped abruptly and locaon acquision ended; we were unable to retrieve his collar or confirm his fate. We would need more data to examine if I-10 is acng as a barrier to mountain lion movements or to examine dispersal in the area in detail; however, our data reaffirm the high mortality risk associated for dispersing individuals.

Fig. 3: An example of 95% ulizaon distribuons (home ranges) from autocorrelated kernel density esmator (AKDE) with 95% confidence intervals based on connuous me movement models for A) a female and B) a male mountain lion (Puma concolor) in the Davis Mountains, Texas, U.S.A.

As expected, resident female and male adults exhibited different movement and space-use paerns. Females moved at slower rates and had smaller home ranges than males (Figure 3), a common theme among large carnivores. From the connuous me movement models, we also obtain esmates of metrics that provide informaon on the general path tortuosity of the animal’s movements and a mescale for the coverage of a constrained area (termed home range crossing me). The models indicated that female and male mountain lions in the study traveled with similar measures of direconality and with similar home range crossing mes. This indicates that the males covered more space by moving faster than females, but crossed the landscape with similar path paerns. The highly rugged terrain that mountain lions select for at broad scales in the area (Dennison et al. 2016) could parally explain the similaries in path direconality. We are further invesgang the movements and space-use of the mountain lions in this wild landscape in West Texas. Our results can potenally help provide insight for migang conflict with humans and for insight into potenal dispersal corridors for mountain lions among the sky islands.

Remotely Wild - Spring 2020 20 SETWG

Literature Cited:

Calabrese JM, Fleming CH, Gurarie E. 2016. ctmm: an R package for analyzing animal relocaon data as a connuous-me stochasc process. Methods in Ecology and Evoluon 7:1124–1132. Dennison CC, Harveson PM, Harveson LA. 2016. Assessing habitat relaonships of mountain lions and their prey in the Davis Mountains, Texas. The Southwestern Naturalist 61:18–27. Fleming CH et al. 2018. Correcng for missing and irregular data in home-range esmaon. Ecological Applicaons 28:1003–1010. Fleming CH, Calabrese JM. 2017. A new kernel density esmator for accurate home-range and species-range area esmaon. Methods in Ecology and Evoluon 8:571–579. Fleming CH, Calabrese JM. 2019. ctmm: connuous-me movement modeling. R package version 0.5.6. . Harveson LA, Route B, Armstrong F, Silvy NJ, Tewes ME. 1999. Trends in populaons of mountain lion in Carlsbad Caverns and Guadalupe Mountains Naonal Parks. The Southwestern Naturalist 44:490–494. Harveson PM, Harveson LA, Hernandez-Sann L, Tewes ME, Silvy NJ, Piman MT. 2012. Characteriscs of two mountain lion Puma concolor populaons in Texas, USA. Wildlife Biology 18:58–66. Holbrook JD, DeYoung RW, Janecka JE, Tewes ME, Honeycu RL, Young JH. 2012. Genec diversity, populaon structure, and movements of mountain lions (Puma concolor) in Texas. Journal of Mammalogy 93:989–1000. Noonan MJ et al. 2019. A comprehensive analysis of autocorrelaon and bias in home range esmaon. Ecological Monographs. Pitman RT, Swanepoel LH, Hunter L, Slotow R, Balme GA. 2015. The importance of refugia, ecological traps and scale for large carnivore management. Biodiversity and Conservaon 24:1975–1987. Piman MT, Guzman GJ, McKinney BP. 1999. Ecology of the mountain lion on the Big Bend Ranch State Park in the Trans-Pecos Region of Texas. Project number 86, TPWD Press. Ausn, Texas, USA. R Core Team. 2019. R: A language and environment for stascal compung. R Foundaon for Stascal Compung, Vienna, Austria. URL hps://www.R-project.org/. Ripple WJ et al. 2014. Status and ecological effects of the world’s largest carnivores. Science 343:1–11. United States Census Bureau. 2018. Quick facts: populaon esmates, July 1, 2018, (V2018); Jeff Davis County, Presidio County, and Brewster County, Texas. Available from hps://www.census.gov/quickfacts/fact/table/ jeffdaviscountytexas,presidiocountytexas,brewstercountytexas,TX# (accessed October 28, 2019). Young JH, Tewes ME, Haines AM, Guzman G, DeMaso SJ. 2010. Survival and mortality of cougars in the Trans-Pecos region. The Southwestern Naturalist 55:411–418.

Remotely Wild - Spring 2020 USFWS 21 SETWG SETWG Spatial Ecology & Telemetry Working Group On the Web at: http://wildlife.org/setwg/

The Spatial Ecology and Telemetry Working Group provides an opportunity for TWS members to address issues of concern to the GIS community and to advance their own skills and understanding of GIS, remote sensing, and telemetry technologies. The Working Group functions as a clearinghouse of information and expertise in the area of GIS, remote sensing, and telemetry for The Wildlife Society Council, TWS sections and chapters, and individual TWS members.

2020 Working Group Executive • Officers Chair — Jeff Jenness, Jenness Enterprises, Flagstaff, AZ [email protected] • Past Chair — James Sheppard, San Diego Zoo Institute for Conservation Research, Escondido, CA [email protected] • Treasurer — Kathy Zeller, Massachusetts Cooperative Fish and Wildlife Research Unit, U of M, Amherst, MA [email protected] • Secretary — Alex Wolf, Scenic Hudson, Inc., Poughkeepsie, NY [email protected]

REMOTELY WILD Spring 2020 – Volume 33 Remotely Wild is a virtual publication issued by the Spatial Ecology and Telemetry Working Group of The Wildlife Society. The newsletter provides information about the working group and its activities, columns and features, information about new technologies, publications and resources of interest to spatially enabled wildlife professionals.

Upcoming Events

2020 Wildlife Society Annual Conference, Louisville, KY: Sep 27- Oct 01, 2020. http://www.twsconference.org

SCGIS Conference, Jul 08 to Jul 10, 2020, Asilomar Conference Grounds, Pacific Grove, CA. http://www.scgis.org/conference

ESA Annual Meeting, August 2-7, 2020, in Salt Lake City, Utah. https://www.esa.org/saltlake/

AniMove summer school 2020, Sept. 7th – 18th, 2020, Max Planck Institute of Animal Behavior: http://animove.org/courses/ animove-2020/

7th International Biologging Symposium, Honolulu, Hawaii, 11 - 16 Oct https://www.bio-logging.net/

Remotely Wild - Spring 2020 22