Prepared for: Natural Resources and Environmental Affairs Division Marine Air Ground Task Force Training Command Twentynine Palms, CA 92278

Transcription

1 RISK OF ATTRACTING PREDATORS FROM HUMAN AND HUMAN-DOG TEAM WILDLIFE SURVEYS Contract Numer W911NF Mary E. Calk, Ph.D. Desert Research Institute 2215 Raggio Parkway Reno, NV (voice/fax) with Jill S. Heaton, Ph.D. University of Nevada, Reno Department of Geography/154 Reno, NV and John C. Sageiel, Ph.D. Desert Research Institute 2215 Raggio Parkway Reno, NV (voice/fax) Prepared for: Natural Resources and Environmental Affairs Division Marine Air Ground Task Force Training Command Twentynine Palms, CA Decemer 6, 2004

2 Tale of Contents Project Overview... 1 Project Overview... 1 Background... 2 Ojective... 4 Methods... 4 Results... 7 Control Plots... 7 Treatment Effects... 7 Temporal Effect Discussion Learning the human scent-food reward relationship Are coyotes or other predators attracted to survey locations? The risk of tortoise predation exclusively from visitation related to human presence A new hypothesis on risk of attracting predators to desert tortoise locations Conclusions References... 26

3 Index of Figures Figure 1. Transect study design. Transects were walked for approximately 50m. At the end of each transect an 8x8m plot was installed. Environmental data were collected, presence of sign recorded, and the area was raked Figure 2. Example plot and data collection. Plot has een surveyed for existing sign (scat and/or tracks). Environmental data were collected and the surrounding area was raked. All data were recorded directly into a data logger attached to a GPS Figure 3. Locations where scat, tracks, or oth scat and tracks were found efore treatment. These are sign recorded at control plots Figure 4. Location where new sign was recorded after treatment Figure 5. Frequency histogram of duration of treatment (x-axis), which ranged from 2 hours to 24 hours 42 minutes. The two treatments were human surveys or human-dog surveys Figure 6. Raked area used to demonstrate the difference etween "pass through" tracks and "investigative" tracks. No tracks are present Figure 7. Example of "pass through" tracks in experimental raked area. Arrow follows direction of travel Figure 8. Demonstration of what "investigative" tracks would look like. Here, the dog stopped to investigate the presence of food and to urinate (arrow on left) Figure 9. Scat dropped at location where vehicle was parked for an extended time period. This scat is atypical and is proaly almost entirely anal gland expulsion

4 Index of Tales Tale 1. Summary of predator sign for control plots. Numer indicates how many plots had scat, tracks, oth scat and tracks or neither scat nor tracks... 7 Tale 2. Summary of predator sign for treatment plots. Numer indicates how many plots had scat, tracks, or neither... 7 Tale 3. Cross-taulation of treatment results with control results. Top row laels correspond to what type of sign was found efore plots were installed at that location. Column lael corresponds to what was recorded at treatement plots Tale 4. Minimum amount of time etween recorded sign after treatment... 11

5 Project Overview The Desert Tortoise (Gopherus agassizii) occurs throughout the Mojave Desert of California, including the Marine Corps Air Ground Comat Center (MCAGCC), Twentynine Palms, CA. A large, multi-agency project is ongoing to test the efficacy of trained scent detector dogs in conducting desert tortoise presence/asence and density/distriution surveys at the National Training Center (NTC) Ft. Irwin. The project descried here contriutes to the larger NTC desert tortoise-k9 project effort. Specifically, we address an issue of concern and suggested y the U.S. Fish and Wildlife Service (FWS) to determine if the use of domestic dog/handler teams to survey for desert tortoises results in attracting native and non-native predators (i.e. coyotes, foxes, feral dogs) at a greater rate than human presence does alone. The ojective was to determine if coyotes, foxes and/or feral/free ranging domestic dogs (here after referred to only as feral dogs) are more attracted to locations where people with dogs conduct field surveys compared to locations where there are only people surveying. There are two elements within this ojective; the first element is to determine simple presence/asence (P/A) as indicated y tracks and/or scat. Presence of coyote, fox and/or feral dog at locations where people and people plus dogs are conducting concentrated work may indicate an elevated interest in that particular location. Defecation or other marking ehavior may indicate a territorial response rather than a predatory reaction. Comparing the amount and type of sign, either tracks or scat, at locations without any human or dog scent (control sites), locations with only human scent (human treatment), and locations with oth human and dog scent (human/dog treatment) will provide asic information as to the degree of relative risk that either survey method poses of drawing the attention of a potential predator. The second element of interest is the length of time etween the human or human/dog presence and the recorded P/A if any, of a potential predator. We did not differentiate etween coyote, fox and/or feral dog, and instead grouped them into the generic category of predator throughout this report. The greater the length of time etween investigation of human or human/dog presence y the coyote or feral dog, the less likely a tortoise is to e in the same location where the survey activity located that tortoise. Thus, the likelihood of an interaction etween tortoise and coyote or feral dog decreases. The comination of oth initial attraction and time to investigation therefore comprises the attraction-risk metric. It is important to clarify that this project was designed and intended to assess the attraction of coyotes, foxes and feral dogs to survey locations and is not intended to determine risk of actual predation to tortoises as a result of either survey method. Because no tortoises were used in this study, the only attraction that could have occurred was to people and/or dogs. The primary fieldwork approach was to monitor the presence of these predators using replicated control plots with human or human plus dog scent. Predator attraction was assessed in terms of actual visitation y recording sign in the form of tracks and feces. The length of time etween installation of the trial plots, where humans or humans/dogs mimic survey data collection, and the first sign of a predator was measured. To date only one study has specifically addressed coyote response to what are termed novelties (Harris & Knowlton 2001) and no formal studies have een conducted to evaluate predation on desert tortoise y coyotes or feral dogs. Coyote wariness of novelties such as cameras or traps is well documented (Winderg 1996,Shivik & Gruver 2002,Sequin et al. 2003). Therefore, this project contriutes valuale information towards understanding potential risks that two - 1 -

6 different survey tools (people and dogs) have of attracting potential predators to desert tortoises. Furthermore, it is of interest to the FWS and the Department of Defense to have an understanding of potential risks of using dogs trained to find desert tortoises in the natural field setting. In terms of roader practical relevance, it is expected that this project will contriute to the creation of standards or guidelines for wildlife detection dogs that the FWS and California Department of Fish and Game may later adopt. Background Desert tortoises occur on military ases throughout the Mojave Desert and are the suject of deate with respect to methods for detection and monitoring, as well as population status and trends. Accurate and reliale methods for detecting individual desert tortoises, determining presence/asence, deriving population estimates, quantifying population status and assessing threats and impacts are not only imperative, ut required for desert tortoise de-listing (Fish and Wildlife Service 1994). Survey methods involve the use of people to locate, approach, and handle target species, including desert tortoises (Anderson et al. 2001,Swann et al. 2002). By locating and going to target species, humans have the potential to catch the attention of and potentially generate interest in that target species location y other species including potential predators. This is called the investigator effect. The investigator effect on predation of reeding ird nests has een studied in reeding ird populations (Bider & Weil 1984,Bety & Gauthier 2001). In controlled field experiments humans were unale to visit nest sites without leaving scent ehind and no difference in predation was found etween sites visited y people only and sites visited y people and dogs (Bety & Gauthier 2001,Donalty & Henke 2001).While the impact of humans and humans with dogs on a leash have een studied in the context of recreational impacts on songirds and mule deer (Miller et al. 2001), data collection on threatened and endangered species requires that people necessarily e involved and as a result the focus on human-wildlife interaction is different in our application from simple periodic disturance to a non-listed species. Without the human interaction, data on individual animals cannot e collected, as direct contact is required to determine size, age, sex, weight and general health of the animal. At the same time, people have not een found to e relialy accurate at finding all size classes, urrows, or scat, regardless of experience or level of training (Freilich & Larue 1998,Anderson et al. 2001). Although precautions are taken to minimize stress and the chance of disease transmission, the presence of people in the desert may attract the attention of potential predators such as coyote, fox or feral dogs (Sequin et al., 2003). Work y (Herman & Valone 2000) evaluated the impact of mammalian predator urine on prey foraging ehavior in the Mojave Desert ut to date no definitive work has een done to document the capacity for humans and humans with dogs to e a canine predator attractant. One study addressed increase in predation y coyotes at artificial watering sources and concluded that predation at these sites was not significantly higher than elsewhere ecause the attraction was for drinking purposes, not predation (DeStefano et al. 2000). The question of learned ehavior in predators in terms of human-reward through indirect scent association is particularly interesting in the desert setting where low relative humidity is thought to correspond to low levels of scent. Recent research has shown that dogs are ale to find desert tortoise under controlled ut natural field conditions and dogs are eing used in iological surveys to locate mammalian scat (Smith et al. 2001,Smith et al. 2003,Wasser et al. 2004). Coyotes, foxes and feral dogs - 2 -

7 are canines and defend territories against intruders such as other canines, certainly within species groups ut also among species. In particular scat, urine, and scent posts are known attractants to coyotes and dogs (Allen et al. 1999). Therefore it is of interest to determine the extent to which the presence of humans, dogs, and human/dog teams attracts coyotes and/or feral dogs, and in particular the lapse time etween human-tortoise interaction and susequent evidence of coyote and/or feral dog investigation of that presence. If coyotes, eing particularly curious yet wary individuals in the case of alpha-animals, (Riley et al. 2003), investigate human presence as readily and as often as human/dog presence then the implications for using human/dog teams as a survey tool for desert tortoise will e that they are unlikely to pose an increased risk to the desert tortoise over the use of human surveyors. Because feral dogs have een reported to harm desert tortoises in the California Mojave Desert, including MCAGCC and on adjacent Bureau of Land Management (BLM) managed lands, we included potential attractive effects of domestic dogs on feral dogs in this study as well. Harassment of tortoises y dogs has not een reported in peer-reviewed scientific literature and to date neither the rate nor extent of damage to tortoises has een rigorously studied. Nevertheless dog harassment is known to occur and is considered a serious threat to tortoises in California. Humans are known to susidize predators, not only at a [human] community level (Boarman 2003) ut also at an individual level, where people drop food, food containers, or other items inadvertently or purposefully while out in the desert. This type of attractant is different from a purely human scent attractant, as predators such as wild or feral canines do not receive a susidy in terms of food or water from the mere presence of a human. Something edile must e left ehind to provide susidy. Coyotes, or in particular feral dogs, will have to have learned to associate human scent with a susidy to ehave in a susidy-seeking manner, which would presumaly e a different ehavioral response from a territorial one. Due to the remoteness of some of the locations selected, we assumed that coyotes would not have learned to associate human scent directly with food or water. Other survey locations (such as Camp Wilson vicinity) are known to support egging or foraging activities in coyotes and feral dogs and thus the human scent-food or water association may e a factor. However, it should e noted that even in uranized areas, coyotes do not typically prefer or develop a predominantly human-susidized diet (Riley et al. 2003). We cannot make the same assumption aout feral dogs ecause some of the feral dogs in the study area may in fact e domesticated pets that rely on their humans for regular meals. Furthermore, research suggests that feral dogs are typically dependent to a large extent on human susidies (Boitani & Ciucci 1995) and as such are less likely to spend as much time seeking food to meet asic life requirements as are native canines. Coyotes have a documented social structure (Gese & Ruff 1997). Alpha coyotes are highly oservant in their territories, which will vary in size depending on resource availaility (Patterson & Messier, 2001). There are typically two alpha coyotes per territory, a male and female, who are a reeding pair (Gese 1998). They tolerate eta coyotes who constitute the pack (Gese 2001). Beta coyotes are typically not as wary nor as oservant. Alpha coyotes were found to oserve humans during the process of installing infrared camera stations with the ojective of photographing coyotes, and later investigated the area, while eta coyotes did not. Alpha coyotes were also found to approach more often when the researchers were - 3 -

8 accompanied y a dog (Sequin et al., 2003). In this research, although coyotes were visile to the researchers intent on capturing photographs of the coyotes, few photographs were actually captured. The researchers reported that coyotes were oth attracted to and wary of, the human presence however, the extent to which the draw was visual versus scent ased was not studied. In our study novelties were not employed as equipment is not typically left in place during or after a desert tortoise has een surveyed although flagging may e left ehind. A novelty is any equipment or gear that an animal would not necessarily encounter on a regular asis and is also new and unfamiliar to an animal. Examples of novelties include a temporary weather station or a remotely triggered camera. As such, any attraction a coyote, fox or feral dog has to the location where collection of data on desert tortoise was simulated must e ased on the coyote s visually ased interest (i.e., must see the survey in process), scent-ased interest, or a comination of the two, ut not due to a persistent non-human attractant such as equipment left ehind in place. Based on research y Sequin et al. (2003), we expected response only from alpha coyotes ut had no way to assess the social status of any individual that left sign. Given the size of the area we surveyed we expected to install plots within territories of many different packs although the numer, size, and distriution of coyote, fox, or dog packs on MCAGCC is not documented. Ojective The ojective was to determine if coyotes, foxes and/or feral dogs are more attracted to locations where people conduct field surveys compared to locations where people and dogs conduct surveys. There are two elements within this ojective; the first element is to determine simple presence/asence as indicated y tracks and/or scat. Presence of coyote, fox, and/or feral dogs at locations where people and people plus dogs are conducting concentrated work, such as taking measurements and iological data of a located desert tortoise may indicate an elevated interest in that particular location. Defecation or other marking ehavior may indicate a territorial response rather than a predatory reaction. The second element was to document the relationship etween the amount of time etween plot installation and visitation y a potential predator. Because the desert tortoise is likely to e a moile target during surveys, the amount of time that lapses etween the tortoise-human interaction and the visitation of a predator to that site is an important factor in quantifying the risk of interaction etween a tortoise and a potential predator due to survey activity and direct interaction. Methods The research was conducted on MCAGCC at Twentynine Palms, CA. All fieldwork was coordinated with Range Control and the Natural Resource and Environmental Affairs (NREA) office. Fieldwork was conducted from Octoer 1, 2004 through Octoer 7, 2004 in oth live-fire ranges (Quackenush and Gays Pass) and non-live fire ranges (Acorn, Sandhill, Gypsum Ridge) (see Appendix 1 - Appendix 3). Survey areas were conducted in ranges known to have coyotes, foxes and/or feral dogs. Field data were collected with Tripod Data Systems (TDS) Solo Software using a Garmin 12XL attached to a TDS Ranger data logger. GPS data were collected in UTM zone 11, NAD83 and exported directly into ESRI native shape file format. All GIS analyses were conducted in ESRI ArcGIS and all statistical - 4 -

9 analyses were conducted using Analyze-It, Splus, and Excel software packages. Fieldwork was designed to simulate field surveys where two individuals walked paired 50 meter transects (Figure 1). At the end of each transect the surveyors installed ~8m 2 plots in which they pretended to take iological data on a desert tortoise (Figure 2). Environmental data were collected: amient surface temperature and relative humidity on the surface. The date and time of day was recorded. Presence of existing sign (scat and/or tracks were recorded) and photographed. This initial data collection served as control plots, representing Surveyor 1 Surveyor 2 Human Human Human + Dog 300 m Human + Dog Human Human Human + Dog Human + Dog Figure 1. Transect study design. Transects were walked for approximately 50m. At the end of each transect an 8x8m plot was installed. Environmental data were collected, presence of sign recorded, and the area was raked. conditions that exist without either humans or dogs. Transects were 300m apart and treatments were paired in opposite directions, with alternating treatment (i.e., human then human and dog). Humans urinated only in plots that were human treatments and likewise dogs urinated and/or defecated in dog plots only. Urination was not regularly prescried, rather, was allowed to occur opportunistically to simulate actual field conditions. The study plots were designed ased on the premise that coyotes may investigate a raked area ut may not walk on the raked surface, ecause it will e novel, or different. For this reason, areas were not raked in a regular pattern although areas were raked to completely surround the main location of concentrated scent. If an animal or a human urinated, that location was raked in a manner to e completely surrounded. The field crew returned to the surveyed plots in increasing time-steps. No survey plot was revisited more than once, as the repeat visit might have increased the interest of the coyote artificially and it is unlikely that repeat visits within a very short time frame (e.g., less than two hours) would occur during an actual desert tortoise survey. The longest time to revisit was approximately 24 hours, as after one day the likelihood of visitation with no visual - 5 -

10 attractant is very low and random visitation on our part would have potentially skewed survey results. Furthermore, the likelihood of a tortoise remaining in the same location for more than one day outside the protection of a urrow during the survey time period is low. Data collected at plot re-visitation included presence of tracks, presence and numer of scat, presence and, if relevant, amount (numer of locations) of urination (if determinale), and any sign from other species such as tracks or scat. The presence of coyotes, foxes, or feral dogs during plot installation was to e noted while conducting the visual surveys. Fisher s Exact test and Chi-squared analyses were run to determine if there was a significant difference in proportions of sign versus no sign for each treatment type. McNemar change test was run to determine if there was an effect from pre-existence of sign in a plot. Figure 2. Example plot and data collection. Plot has een surveyed for existing sign (scat and/or tracks). Environmental data were collected and the surrounding area was raked. All data were recorded directly into a data logger attached to a GPS

11 Results Control Plots One hundred and ninety eight plots were installed over the course of six days, of which 198 were controls, 95 were human/dog treatments, and 103 were human only treatments. The control plots represent the presence or asence of predator sign without human or human/dog attractant. Therefore, each treatment plot was also a control plot. Tale 1 summarizes predator sign for control plots, which is the amount and type of sign that was already present where treatment plots were installed. Figure 3 shows where sign was found at control plots, efore treatment. Some type of predator sign was present in the form of tracks and/or scat during field surveys in 36% of the control plots. More often than not that sign was only scat (28% scat), ut a human oserver encountered scat and/or scat with tracks in 8% of the control plots. Tale 1. Summary of predator sign for control plots. Numer indicates how many plots had scat, tracks, oth scat and tracks or neither scat nor tracks. Treatment Effects Type Control Scat 9 Tracks 56 Both 7 Neither 126 Scat was encountered in one treatment plot and tracks were recorded in an additional nine different treatment plots (Figure 4) for a total of eight positive responses to pooled treatments. The scat encounter was in a human only plot and six of the nine recorded track incidents were in human only plots. Predator sign (scat or tracks) was recorded in six of 103 human treatment plots for a rate of 6% human-predator interaction possiility. Predator tracks were recorded in three dog treatment plots. Three out of 95 dog plots had recorded predator sign for a rate of 3% human-predator interaction possiility. Appendix 4 and Appendix 5 show locations where sign was recorded in treatment plots. These results are summarized in Tale 2. Chi-squared analysis indicated that there was no statistically significant difference etween using humans or using humans with dogs (n = 198, X 2 = 1.31, p = ) and whether or not there was presence of new sign. Fisher s Exact test results showed no significant difference in the proportion of sign compared to no sign for treatment plots (n = 198, p = ). Tale 2. Summary of predator sign for treatment plots. Numer indicates how many plots had scat, tracks, or neither. Type Human only Human and Dog Scat 1 0 Tracks 5 3 Neither

12 QUACKENBUSH LAKE qp qp EMERSON LAKE qp JI JI GYPSUM RIDGE JI ACORN qpqp JI JI JI TORTOISE AREA qp JI JI SANDHILL JI Legend PRED_CNTRL qp JI oth sca t trax Figure 3. Locations where scat, tracks, or oth scat and tracks were found efore treatment. These are sign recorded at control plots

13 JI QUACKENBUSH LAKE EMERSON LAKE GYPSUM RIDGE ACORN TORTOISE AREA SANDHILL Legend Recorded Predator Sign JI sca t tracks Figure 4. Location where new sign was recorded after treatment

14 Tale 3 summarizes the relationship etween what sign was found at control plots and what was found after treatment at those same locations. For those treatment plots with recorded new predator tracks, four had tracks at their location prior to installation and two had oth scat and tracks recorded. Three plots with new tracks recorded had no sign of predators efore eing installed. A human urinated in the one plot where fresh scat was recorded in a treatment plot. The scat was dropped within centimeters of the urination. No sign was recorded at this location efore the plot was installed. Fifty-two treatment plots had no new predator sign where tracks were recorded during plot installation. Nine treatment plots had no new predator sign where scat was recorded during plot installation. Five treatment plots had no new predator sign were oth scat and tracks had een previously recorded. Finally, 123 plots had no new sign recorded and had no sign prior to installation. McNemar change test result (p < ) indicated no significant difference etween plots that had sign prior to the treatment and those that did not have sign prior to treatment. Tale 3. Cross-taulation of treatment results with control results. Top row laels correspond to what type of sign was found efore plots were installed at that location. Column lael corresponds to what was recorded at treatement plots. Control Plots Type Scat Tracks Scat and Tracks None Scat Tracks No new sign Temporal Effect Plots were treated and then left alone and undistured for differing amounts of time to assess the immediacy of any potential predator-treatment interaction. The treatment time period ranged etween 2 hours and 24 hours, 42 minutes total. Figure 5 shows the treatment time spread as a frequency histogram. Tracks were recorded in treatment plots from 12hrs 33mins to 18hrs 47mins after installation. The scat in the treatment plot was recorded almost 16hrs after treatment (Tale 4). All treatment plots with new sign recorded included the night hours in their time duration. No plots that did not include the dusk to night hours had new sign recorded. Only plots that were left undistured overnight were recorded as having sign

15 Figure 5. Frequency histogram of duration of treatment (x-axis), which ranged from 2 hours to 24 hours 42 minutes. The two treatments were human surveys or human-dog surveys. Tale 4. Minimum amount of time etween recorded sign after treatment. Sign Minimum time to recorded sign Scat 16 hrs Tracks 12 hrs 33 min Thirty-eight treatment plots had human urine or dog urine/feces as part of the treatment. Three treatment plots that recorded new sign were urinated in y humans while none of the plots where dogs either urinated or defecated had new sign recorded. Other sign recorded at these locations were either military, such as ordnance or other training related ojects, or none recorded. No new sign was found in locations that were noted as having food related paraphernalia such as drink containers or food packages. Discussion Learning the human scent-food reward relationship The concern that the use of dogs to survey for desert tortoise will increase predation y coyote, fox, or feral dog parallels similar concerns in ornithological studies where nest predation is thought to increase due to the oserver effect. It is assumed that when humans visit a nest, predators locate the human scent, follow the scent to the nest, and gain a reward, such as eggs as a food source, for doing so. The argument for this logic is ased largely in the fact that most mammalian predators have well-developed olfactory ailities (Donalty & Henke 2001). However, whether or not an animal has a well-developed olfactory sense does not necessarily correlate to exploitation of all possile uses of that sense. For this reason there are several critical logical steps missing from the assumed progression of human scent to predation of any species or nest, e it ird or tortoise or other. We discuss why this assumed learning is a false assumption. First, animals learn through operant conditioning and through aversive conditioning (Beckmann & Berger 2003). Operant conditioning occurs in the sequential presentation of stimulus-reward and stimulus includes events. Aversive conditioning is the learning of a

16 direct negative response to a stimulus. Aversive conditioning has een studied as a possile predator control method to protect domestic livestock where animals are fed meat tainted with a sustance that then makes them temporarily ill. The animal learns the association etween that type of meat and the illness and avoids it. It is important to note that the animal does not learn the association etween human scent, where humans handled and placed the meat, and the illness. Coyotes are successful at existing in uran areas and may e susidized as ears are in some national parks or other areas without ear-proof trash receptacles. These animals may e susidized and may e desensitized to humans, ut are not necessarily attracted to human scent. If human scent were the direct learned cue, then coyotes would walk up to humans at every opportunity with the expectation of a food reward. This is not a typical ehavior pattern. In the natural field setting, an event happens, such as a human tosses food towards an inquisitive coyote, and the animal is rewarded. Such ehavior y humans has taught coyotes to directly approach vehicles in Joshua Tree National Park among others. This stimulusreward concept is the asis for the method employed y aquariums to train marine mammals to perform complex tricks. For an animal to learn a correlative relationship etween stimulusreward, or event-reward, there must e a direct link etween the two. If a coyote happens to find an anthropogenic food source at a location, such as along a road where a vehicle has recently passed, the coyote is rewarded at that location. It is incorrect to assume that the coyote has in that instance learned the human-food association ecause first, there is no human present, and second, there are a host of other scents to which the association could e tagged such as vehicular emission, ammunition, gunpowder, hydrocaron from fuels, etc. Most of those scents would e stronger than human scent, potentially have a longer residence time, and thus e more likely to elicit the reward association. However, the most likely association in this example would e a location association ecause there was not a particular event that triggered a learned response. The coyote happened upon the food y chance. If more food is found on or adjacent to other roads or along different stretches of the same road, the coyote will generalize the reward setting to roads, not to people. This is ecause if coyotes repeatedly find food (reward) on or near roads, they are likely to revisit these locations, ecause there is some proaility of finding food. As the rate of visitation increases, assuming food continues to fall out of vehicles on roads, the rate of reward increases as well. The learning of place-reward is strengthened and humans are not a direct link in that learned association. For humans to e a direct link in that association coyotes would have to have direct interaction with a human. We did not experience any instances where coyotes approached our vehicle or us. We were arked at on one occasion during daylight hours, ut we could not see the animal to determine if it were a coyote or a dog, or if there were more than one animal. This event occurred near the east oundary of Range Range road and Quackenush. Are coyotes or other predators attracted to survey locations? Domestic canines are known to investigate and re-mark locations where another domestic canine has previously marked (Bekoff 2001). It is likely that feral dogs ehave similarly as do native canines and would thus e attracted to locations where other domestic canines, such as dogs used in desert tortoise surveys, had een. This is not what the data we collected showed, however, as we did not oserve any investigation of any site y a feral or domestic

17 dog. Plots in the non live-fire ranges were in areas where feral and pet dogs have een seen in the past and dog tracks were found in the area unrelated to our plots. The results from this study indicate that predators are not attracted to survey locations qualified with the statement that the term survey does not include leaving ehind equipment or remaining stationary in one location overnight. Coyotes engage in more investigative ehavior in low-use areas of their home range when a potential food source is availale as a draw (Hein & Andelt 1996) or when the food source is particularly large (Kamler et al. 2004). In familiar environments, such as in a coyote s home range, the coyote is more likely to e wary of visual cues than olfactory cues (Winderg 1996). This lends support to the idea that coyotes and other predators need to see a structure or see the humans doing activity to e attracted or to investigate when food is not directly involved. Based on this evidence it is unlikely that the mere presence of human scent or dog scent is enough to draw a coyote and our analysis supports this premise. Coyotes are repeatedly cited in peer-reviewed literature as eing wary and curious at the same time and the two motivations are conflicting from a ehavioral standpoint. One cannot simultaneously act on attraction and repulsion. For this reason a coyote may e drawn to a location where a survey team is camped or parked for a significant amount of time, and potentially could e attracted to a location with a desert tortoise if a vehicle or other structure co-existed, ut there is little evidence to support the idea that humans or humans with dogs conducting field surveys will attract predators to a transitory data collection location. Furthermore, predators may have indeed een attracted to a treatment area due to the presence of human-related scent, ut not interested enough to define the exact pinpoint location of where we stood collecting data. Perhaps the fact that food was not intermixed in the treatment plot scent had an impact on the extent to which a predator investigated that location. To test the idea that the track patterns seen and descried as pass through were consistent with the ehavior as descried, we conducted an experiment with the research dogs. An area was raked in the same manner as done in the study plots (Figure 6) and one dog was called y the handler so that it moved through the raked area directly to the handler. In this manner the pass through pattern was simulated (Figure 7). These tracks were consistent with those seen in sites descried as pass through. In contrast, we placed a small amount of food in the middle of a raked area and allowed one dog to investigate that plot. The pattern left in the food investigation plot was sustantially different from the pattern that resulted from simple traveling through the area (Figure 8). The food investigation pattern had many footprints close together with no clear direction of travel. For this reason, we suggest that the pass through pattern seen is consistent with the travel of an animal that did not investigate the area immediately surrounding the data gathering activity. Interpreting the data in the strictest sense, coyotes, foxes and/or feral dogs were more attracted to locations where people conduct field surveys than locations where people and dogs conduct surveys. In fact we were twice as likely to find sign at human only plots than at human and dog plots. This relative proportion is not statistically significantly different ut is worth discussing from a practical and risk assessment perspective. We say strictest sense ecause we find the track evidence inconclusive regarding whether or not predators were in fact attracted to sites with rare exception of two sites, which we will discuss later. None of

18 the tracks encountered demonstrated what would e considered an investigatory pattern where an animal stopped at a minimum and shuffled within the raked area at a maximum. To the contrary, tracks recorded in raked areas were what we deem pass through, that is continuous stride of unroken cadence entering from one side of the raked area and exiting the opposite side. However, we can neither rule out that the predator passed over the raked area and then stopped to investigate from the edge of the raked area. In many of the treatment locations the surrounding area was too hard to sustain a track and for that reason we cannot conclude for certain that the predators were or were not necessarily attracted to the treatment. Figure 6. Raked area used to demonstrate the difference etween "pass through" tracks and "investigative" tracks. No tracks are present. Finally, the potential attraction of a predator to human and or dog scent from a scent theory perspective should e discussed. Coyotes are attracted y fatty acids, which are found naturally in scat (Howard et al. 2002). Fatty acids are thought to e the main compounds that contriute to the effectiveness of marking and are used as a means of social communication even in animals such as the desert tortoise (Rose 1970,Alerts et al. 1994,Bulova 1997). It is unlikely that a human or a dog walking or performing survey tasks leaves fatty acids at an area unless they defecate and even then, there is no evidence that humans and canines evolved in a competitive fashion, as intraspecific species competition. We have one data point where a coyote marked directly over where a human marked, ut this is one data point and could not e considered a trend. There was one additional instance when a coyote

19 marked the location where humans and dogs had spent several hours in one place, which is descried elow. For this reason the scent of a human or the scent of a human and a dog comined proaly does not serve as a natural attractant to a canine predator. In fact, there is evidence that the presence or scent of domestic dogs would deter coyotes from investigating (Bider & Weil 1984,Andelt 1999,Kamler et al. 2003). Figure 7. Example of "pass through" tracks in experimental raked area. Arrow follows direction of travel. The risk of tortoise predation exclusively from visitation related to human presence People survey for desert tortoises during the daylight hours, focusing efforts at dawn and dusk when tortoises are active and there is enough light for humans to see. For this reason we were particularly interested in capturing activity that would occur at these time periods when the chance of humans encountering tortoises are presumaly greatest. No interactions were recorded at treatment plots that were not left undistured overnight. Based on this result it would appear that there is zero risk of predators eing attracted to a given location during daylight hours, the typical time frame when humans conduct desert tortoise surveys. We cannot say that the risk of predator-tortoise interaction is necessarily zero during this same time frame ecause we do not know the additional attraction that a tortoise may or may not add to the mix. However, we can say that humans alone and that humans with dogs are not enough to attract a feral dog, coyote, or fox during daylight hours

20 Figure 8. Demonstration of what "investigative" tracks would look like. Here, the dog stopped to investigate the presence of food and to urinate (arrow on left). Coyotes are opportunistic foragers and the majority of their diet is small mammals and other animal matter although vegetation does comprise a small part of their diet (Sperry 1941,Hernandez et al. 2002). There have not een studies quantifying the relative proportion of desert tortoise in coyote, fox, or feral dog diet and therefore we cannot quantify the actual risk of death from human caused predator-tortoise interaction from field surveys. We simply do not know how often or under what conditions a predator will kill and eat a tortoise. MCAGCC has reported tortoises that have een harassed y dogs, although there is no evidence to rule out the possiility that the perpetrator was a coyote, including tortoises with gular horns chewed off, ut these tortoises are not killed in the process and continue to survive eyond the initial attack y dogs. We can make a conservative, or worst-case scenario, ased on certain assumptions. If we assume that 1) predators track humans to tortoises at the rate of return we found in this study; 2) a tortoise will remain in the location where it was found during the survey up to 24 hours and 3) a predator will always eat a tortoise upon finding one then we can calculate the risk of predation ased on human and human/dog surveys. The return rate of predators to human only sites was 6%. Therefore under assumptions 2 and 3 the risk of tortoise predation is 6% from human surveys. Six out of every 100 tortoises found during human surveys would e expected to e predated. Likewise, when a dog was used in the surveys the predator response

21 rate was 3% and thus the risk to tortoises from dog surveys drops y 50%. Three out of every 100 tortoises encountered during dog surveys would e expected to e predated. It is important to reiterate that the difference etween human and human-dog surveys were not statistically significant and also that the assumptions stated aove do not represent realistic conditions in the field. Therefore the 6% and 3% risk to tortoises should e viewed as oth high and the worst-case scenario. The fact that we did not always find sign does not preclude the presence of predators in the immediate vicinity. Sign may have deteriorated, een lown away, covered, eaten, or otherwise removed or hidden. Residence time of scat in the desert has een shown to e very short in some cases as few as four days (Alerts et al. 1994,Sanchez et al. 2004) and we do not know what the local residence time for scat persistence is at MCAGCC. However, ased on the work of Sanchez et al., (2004) it is likely that the scats we located were on the order of a week or less due to the prevalence of kangaroo rat urrows. Also we do not know the age of tracks, whose persistence is very much dependent on environmental conditions such as wind velocity, recent rains and surface type. A new hypothesis on risk of attracting predators to desert tortoise locations There was another marking incident that occurred ut that did not go into the analysis ecause it occurred outside of the trial protocol. We had to park several hours over the dusk into nightfall time range at a location near the intersection of Range Range Road and Quackenush on the northeast corner. During that time we were active in the immediate vicinity of the vehicle. We also allowed the dogs to spend a short amount of time within 6 of the vehicle on leash. After dark coyotes egan to howl from Range Range and from the area west of the vehicle. We left the area etween one and two hours after dark. The next morning we revisited the site. At least five different animals were identified from fresh tracks in the road and in the adjacent erm. One animal defecated in the center of the road at a location that did not correspond with where the dogs had walked ut was very close to where our vehicle had een parked. This location was off the front-right side of our vehicle where the people had spent more time without the dogs. The scat was very loose to the point of holding no shape, was very dark rown, almost lack, and contained short red-rown hairs (Figure 9). The odor was very musky to the human nose and much stronger than any other scat collected. There also appeared to e a paw print in the scat. The literature on coyote ehavior is sometimes contradictory, ut for the most part consistent in terms of the extent to which coyotes are oservant and investigative within their home ranges. Investigatory ehavior has een reported to occur with variaility depending on the study location and in particular the social status of particular animals. Based on the literature and on the results of this study, we suggest that the risk of attracting predators such as coyote from either human or the use of dogs depends on a comination of factors. The first is that the coyotes must see the humans and thus e aware that humans are present. Scent of humans alone is not enough to attract a predator with the possile exception of human urination. The second factor is that the humans must e in place for a certain amount of time, proaly more than a few minutes. The third factor that would dictate predator attraction to a specific location is that some type of equipment or marking must remain in place to continue to e an attractant at that location. Finally, there is likely a relationship etween site visitation y a

22 predator and time of day, where there would e a higher likelihood of visitation after dark than during daylight hours, at least where coyotes are not desensitized to humans. Coyote home range varies with resource availaility (Patterson & Messier 2001,Servin et al. 2003,Wasser et al. 2004) so if we put all the evidence together, from this study and from the literature, we can report with confidence that we did not find a significant risk of attracting predators from the use of dogs. Even if there was a significant risk to tortoises from predators y using humans or humans with dogs, there is no way to know predator population details in a survey area, such as population structure, size and individuals status. Surveyors would have little means to pinpoint the risk, which would e expected to change ased on population and individual animals present in an area, efore conducting a survey. Population dynamics of predators are going to e highly variale and making predictions aout what any one individual animal might do within a population is guesswork. Figure 9. Scat dropped at location where vehicle was parked for an extended time period. This scat is atypical and is proaly almost entirely anal gland expulsion. Conclusions We assessed the degree to which coyotes, foxes and feral dogs are attracted to human scent, and to human and dog scent. We found predator sign in plots with human scent twice as often as in plots with oth human and dog scent, although the overall response to any presence of humans and/or dogs did not differ statistically significantly etween the two treatments. In two separate instances coyotes exhiited territorial response to human scent whereas there was no territorial response to dog scent, even when there was dog defecation or urination. There may e an aversion effect due to competition etween coyotes or foxes and dogs, which could reduce the risk to tortoises during field surveys that employ working domestic dogs. Because there are no statistics quantifying the rate of predation of tortoises y

23 predators we cannot assign a realistic risk quotient to tortoises from surveys, either under existing human ased methods or from the using dogs. However, under some asic and extreme assumptions, we determined that tortoises face a 6% chance of encounter with a predator with human survey efforts and a 3% chance of encounter with a predator if surveyed with humans and dogs comined. Our overall conclusion from this work is that in a setting where coyotes, foxes, and feral dogs are susidized ut retain a healthy wariness or fear of direct human contact, the contact risk to tortoises with a predator is very low to none, depending on the moility of the desert tortoise. Where predators have learned the humanfood reward through operant conditioning, this premise may not hold true and we hypothesize the risk would increase. Where predators have little to no interaction with humans or dogs, we expect the risk of interaction from either survey method to e even lower than we report for this project. The results of this work will support the next phase of research that will egin in 2005 funded y the US Army (National Training Center at Fort Irwin) evaluating the use of dogs to survey for desert tortoise relative to human ased surveys. As such, the nature and extent of the research is important for all desert tortoise surveys and tortoise research in addition to contriuting to a very limited ody of knowledge aout predation risk in the scientific literature at present. More research is needed efore dogs can e deployed for survey missions, and the results of the proposed project are an important and significant step towards that effort. Future research efforts on the Marine Corps Base at Twentynine Palms would contriute to oth the dog research efforts and more importantly, to the ody of knowledge aout the status of desert tortoise populations on military ases in the Mojave Desert. At present MCAGCC, the NTC Ft. Irwin, and Edwards AFB have supported or are in the planning process for employing the use of certified desert tortoise dogs. Should the dog teams fielded in 2005 at the NTC Ft. Irwin e proven safe, effective, and reliale, they will e a valuale research and monitoring tool to investigate a numer of questions aout desert tortoise iology. In particular, etter aundance estimates may e generated, population structure y age and size class may e etter determined, and aseline information on dispersion and recruitment may e acquired. Such surveys done on MCAGCC would complement future work eing planned y Edwards AFB and would dovetail with the 2005 research for the NTC Ft. Irwin to create a regional tortoise aundance distriution. No research has een done yet on the capailities of dogs to locate uried tortoise eggs and this focus might e considered if dogs are proven to e a reliale survey tool in the field setting. The capailities of desert tortoise dogs at night likewise have not yet een evaluated. Benefits from nighttime surveys would provide ehavioral information aout tortoise activity during the hours when humans typically do not survey

24 SUNSHINE PEAK LAVIC LAKE GAYS PASS RAINBOW CANYON MAUMEE MINE BLACK TOP NOBLE PASS QUACKENBUSH LAKE LAVA LEAD MOUNTAIN EMERSON LAKE DELTA GYPSUM RIDGERANGE ACORN TORTOISE AREA AMERICA MINE BULLION PROSPECT CLEGHORN PASS SANDHILL WEST EAST MAINSIDE Plot Location Appendix 1. Study area and plot locations.

25 BLACK TOP MAUMEE MINE GAYS PASS RAINBOW CANYON EMERSON LAKE QUACKENBUSH LAKE NOBLE PASS RANGE dog people Appendix 2. Plot locations in live fire ranges Quackenush and Gays Pass.

26 QUACKENBUSH LAKE DELTA EMERSON LAKE ACORN TORTOISE AREA SANDHILL GYPSUM RIDGE WEST RANGE dog people Appendix 3. Transects in non-live fire ranges Gypsum Ridge, Acorn, and Sandhill.

27 QUACKENBUSH LAKE JI EMERSON LAKE Recorded Predator Sign JI scat tracks Appendix 4. Location where tracks and scat were recorded after treatment in live fire range Quackenush.

28 EMERSON LAKE GYPSUM RIDGE ACORN TORTOISE AREA Recorded Predator Sign tracks Appendix 5. Location of tracks found in non live fire ranges after treatment.