Patterns of fine-scale animal mobility within heterogeneous landscapes have significant impact on foraging and reproductive success, inter and intra-specific competition, creation and persistence of meta-populations, gene flow and structure, and propagation of infectious diseases [1]–[6]. Mobility depends on a complex repertoire of biological, behavioral and environmental processes including animal fitness, habitat selection, dispersion, foraging, social interactions, predator aversion, and reproduction, which may occur across multiple spatial and temporal scales [7]–[10]. Within species, fine-scale mobility patterns vary with resource availability or habitat loss [11], [12].

At fine spatial scales (i.e., landscape level), animal mobility has historically been quantified by direct observation, mark-recapture studies or radio-telemetry [13]–[15]. Direct observations and mark-recapture studies are labor intensive, affected by observer bias, and not suitable for some animal systems. Radio-telemetry, which relies on animals wearing radio transmitters overcomes some of these limitations; however, these systems still require close range observations that can influence animal behavior, limit the number of animals studied, and frequency of data captured [16], [17].

Global positioning systems (GPS) determine the location of an individual on or above the Earth's surface 24 hours a day with a high level of accuracy [18]. The large size of GPS collars has limited GPS telemetry studies to large animals with few individuals [19]–[21]. However, recent reductions in the size and cost of GPS telemetry devices now make them attractive for ecological, biological and epidemiological studies of the interplay between a larger number of smaller wild and domesticated animals, their local environment, and spread of disease.

Domesticated animals play an important role in the persistence, amplification and transmission of zoonotic pathogens and serve as sources of pathogen spillover to endangered wildlife populations, especially when anthropogenic activities lead to habitat overlap between domesticated animals and wildlife [22], [23]. Epizootic diseases such as Chagas and Nipah virus infections cause morbidity and mortality in humans and wildlife; intervention strategies are hampered by the persistence of animal reservoirs, with recursive mobility and pathogen amplification capability [24], [25]. Despite the critical role that domesticated animals play in disease dynamics, a lack of suitable tools to continuously track animal mobility has hindered the study of mechanisms driving zoonotic spillover and disease emergence.

Empirical evidence indicates that the endangered chimpanzee (Pan troglodytes schweinfurthii) population of Gombe National Park, Tanzania, is infected with enteric parasites associated with humans and domesticated animals [26]. They have experienced SIVcpz-associated morbidity [27], which may be complicated by Cryptosporidium co-infection, as seen in human HIV-Cryptosporidium co-infections [28]. Cryptosporidium is a zoonotic gastrointestinal parasite of a wide range of vertebrates, highlighting its transmission potential between host species [29], from ingestion of contaminated food or water [30]. Cryptosporidium is common in ruminants and completes its life cycle in the animal's intestines, shedding high numbers of infectious oocysts into the terrestrial environment [31]. Despite these health threats, few studies have examined the ecology of this pathogen in rural tropical forest systems characterized by high rates of overlap among humans, domesticated animals, and wildlife [32].

The present study aimed to 1) assess the utility of GPS telemetry to capture the mobility patterns of domesticated animals (dogs, goats and sheep) using portable GPS data-loggers and 2) establish putative Cryptosporidium transmission routes by determining the extent of infected domesticated animal overlap with chimpanzees.

Twenty-five households (96.1% of all contacted households) participated in the study. Sample sizes were determined as the number of animals collared over the estimated population size for each species. We collared 8 dogs (100% of village population), 2 sheep (20% of population), and 27 goats (18% of population). Longitudinal data collection was possible for all sheep, 75% of goats and 37.5% of dogs. We were unable to resample 3 individuals due to death (1 goat and 2 dogs), inability to recapture the animal (1 dog) or unavailability of household occupants (1 goat). A single household declined enrollment due to concern that their animal (goat) may get stuck in grasses. Only one GPS collar (4%) was never retrieved due to the animal (dog) being reported stolen.

In total, 123 unique animal readings were captured by GPS. Eight (6.5%) generated <24 hrs of data (range 0:00–22:00 hrs) due to programming error and were excluded from evaluation. The GPS units maintained an average battery life of 92:00 hrs (range: 27:20–123:30 hrs; (SD) = 16:40). Battery life was not affected by season with an average use of 95:50 hrs (range: 27:20–134:50) in dry season versus 89:60 hrs (range 37:40–123:30) during the wet season (p-value  = 0.165). If the average unit held a battery life of 92 hours, then this could generate 2,760 signals per collar rotation given that the units were programmed to capture a point every two minutes. The average number of GPS points captured was 1,634 (60% coverage). Fewer signal points were generated during the wet season (1,501 points) versus dry (1,891), (independent Student's t- test, p-value 0.002).

Five goats (18.5%) and two sheep (100%) were positive for Cryptosporidium. All dogs were negative for the parasite. Infection status was not determined to affect the daily median distance of sheep (Mann-Whitney U = 31, p-value = 0.91) or goats (Mann-Whitney U = 561, p-value = 0.95) from their domicile (Table 1). Goats moved significantly further from home during the wet versus dry season (U = 8,935, p-value <0.001) (Table 1). Small sample sizes prevented us to test whether such seasonal variation was observed in sheep or dogs. There was variation in the distance that dogs moved on a daily basis; one individual dog traveled approximately 4,500 m from its residence in the dry season compared to the canine daily median distance of 24.5 m (IQR: 15–4,565 m). Dogs also had the greatest daily mean maximum distance traveled at 1,876 m in the dry season compared to 589 m for goats and 471 m for sheep (Figure 2).

All species moved into the national park with the median distance of introgression less than 50 m (dogs: median distance 37 m, IQR 20–80; goats median distance 20 m, IQR 11–39; sheep median distance 50 m, IQR 23–196). Of the three species, goats spent the most time in the park in the dry (35% of GPS points collected) and wet (35%) seasons (Figure 3; p-value <0.0001 for both seasons) compared to dogs (≤2% of GPS points) and sheep (≤1%). Outside the park, the crop fields were commonly visited by dogs and goats; in the dry season, 7% of GPS points were from seven goats that visited or moved through crop fields compared to 10 goats (14%) in the wet season. Four dogs (two per season) moved into these fields during the dry (2%) and wet (1%) seasons.

Goats infected with Cryptosporidium did not have different ranging patterns from those that were not infected (X² = 0.157, df = 1 p-value 0.691). The aggregated GPS data showed that of the 5 goats with a specimen yielding Cryptosporidium, infection status did not alter ranging behavior; three goats were found to spend time in the park but not the crops fields while either infected or uninfected. A fourth goat, though Cryptosporidium positive in the dry season, maintained a ranging pattern in both seasons that included the park and crop fields. One infected goat was not found in either location. Infected goats made up 40% (dry season) and 25% (wet) of GPS points of all goats in the crop fields and contributed 2.7% (dry) and 5.6% (wet) of all goats points in the park (Figure 3). Small sample size made it difficult to discern significance of sheep ranging patterns based upon infection status. Both sheep collared were positive for Cryptosporidium in the wet season only. One sheep moved into the park during the wet season but not the dry season. The other infected sheep stayed within the village perimeter. Neither sheep visited the crop fields. The prevalence of Cryptosporidium in the Mitumba and Kasekela chimpanzee communities was 15.4% and 20.6% but the difference between groups was not significant (Fisher's exact test (two-tailed) p-value 0.7660). Rates of chimpanzee crop raiding are unknown since monitoring does not occur outside park boundaries but villager sightings of Mitumba chimpanzees raiding the crop fields occur year round with greater frequency reported in the dry season.

The human-wildlife interface is permeable and regularly altered by relatively unknown ecological and epidemiological dynamics. Our study demonstrates that portable GPS technology allows rapid and concurrent characterization of fine-scale movement from multiple individuals in a population. This technology, in combination with clinical data, can also be used to identify the spatial arrangement of those individuals infected with an etiologic agent at a specific point in time, and determine whether their infection status may alter fine scale movement processes and/or be used to predict probabilities of disease spread over time. The GPS unit cost, battery life, durability, size and weight made it suitable for field deployment. The affordability of the unit ($50 USD each) allowed for tracking 42 animals (up to 12 animals per exchange) simultaneously for two consecutive 96-hour periods. Cost is a limiting factor when considering study design and number of individuals to sample, and GPS data-loggers like the ones used in this study can provide a reliable way for tracking fine-scale local movements of domesticated animals.

The size (L/W/H in mm; 44.5×28.5×13) and weight (20 g) of the units minimized the likelihood of influencing animal movement and behavior and allows for the tracking of smaller animals. The durable, water resistant properties of the unit made it suitable for tracking during excessive rainfall. Strapping the unit to an adjustable, clip-on collar minimized investigator contact with study animals, thereby reducing stress. The low refusal rate by households invited to participate in the study indicated limited apprehension to the units and associated data collection. We collected 117 unique animal readings representing over 10,500 hours of animal mobility data. The only performance difference detected was the collection of fewer GPS points during the wet season, potentially due to factors affecting satellite signal loss such as greater canopy cover, excessive rainfall, or atmospheric conditions [39].

The GPS data provide unique insights on the patterns of domesticated animal mobility and potential interaction with chimpanzees within the park and crop-raiding areas. Our study suggests that goats, due to their high overlap with chimpanzee habitats, are the most likely spillover host and that crop-raiding areas are the potential Cryptosporidium spillover hotspot into chimpanzee populations. The frequency of habitat overlap during the dry season is compelling as chimpanzees are more likely to raid crops as they seek out alternative food sources. Crop raiding is a frequently reported behavior among primates [40]. A Nigerian study reported that crop-raiding baboons were more likely to harbor the anthropogenic parasite Balantidium coli as compared to their less frequent raiding counterparts [41]. Interestingly, Gillespie et al. [26] found that B. coli only occurred in Mitumba chimpanzees and that overall chimpanzee parasite diversity was higher for Mitumba chimpanzees compared to Kasekela chimpanzees. During the study period, it was also reported that villagers may spread animal feces on crops as a natural deterrent (I. Lipende, personal communication), which has the potential to indirectly perpetuate the spread of infection in this identified environmental hotspot.

Cryptosporidium was detected in both Mitumba and Kasekela but there was no significant difference in the prevalence of Cryptosporidium between groups. Although not statistically significant, it was surprising to find a moderate prevalence of Cryptosporidium in Kasekela community as compared to Mitumba. The Kasekela chimpanzees have less interaction with humans and no interaction with livestock as compared to the Mitumba community whose natural border with Mwamgongo village places them in contact with human and animal activities and at greater risk for zoonotic infections. These results suggest that different transmission cycles are operating in these groups. Additional molecular characterization of the positive samples is underway to determine the species of Cryptosporidium in this system to assess potential transmission pathways and sources for exposure.

Cryptosporidium is of global concern [28] and capable of surviving in the environment for long periods. The parasite has been detected in other habituated primates. A bovine genotype was isolated from gorillas and people living in and around Bwindi Impenetrable National Park, Uganda [42], [43]. Salyer et al. [44] examined the molecular epidemiology of Cryptosporidium in wild primates, people and livestock in and around Kibale National Park in Uganda and found red colobus (Procolobus badius) and black-and-white colobus (Colobus guereza) infected with Cryptosporidium parvum/C. hominis that resembled that of people and livestock at the forest edge, while red colobus in the forest interior were infected with a divergent subclade, suggesting the possibility of separate zoonotic and sylvatic cycles. These findings suggest that zoonotic transmission of Cryptosporidium can be frequent and occur with ease in tropical settings where people, livestock, and wild primates overlap.

Sheep were positive for Cryptosporidium, but their movement was predominantly restricted to the village, limiting their probability of contact with chimpanzees. It is of note that only two of the village sheep were tracked and their mobility patterns may not represent movement of all sheep in the village. Although the dogs tracked in this study were all negative for Cryptosporidium, they demonstrated substantial overlap with chimpanzees and may present other zoonotic risks to chimpanzees, such as rabies, foot and mouth disease and parasitic worms.

The findings from this research have local and broad implications. Although there was evidence of animal mobility into the park, the few animal points beyond the forest edge indicate limited park intrusion. Rather, the area where chimpanzees are known to raid crops was a potential hot spot for pathogen spillover due to high visitation rates by domestic animals and the dispersion of potentially infected animal feces. Application of animal feces to crops may perpetuate the transmission of foodborne pathogens [45]. Pathogen introductions into the environment can occur in both the dry and wet seasons but actual spillover to chimpanzees was more likely to occur in the dry season.

The focus of conservation strategies traditionally has been on wildlife corridors to minimize habitat fragmentation and risk for population isolation [46]. These data suggest a need to monitor borders and edges, in wildlife management and park design. In this case, the application of integrated GPS/GIS technology was capable of identifying a novel area along the Mwamgongo/Gombe border most at risk from human and animal use that may impact the sustainability of vulnerable wildlife populations.