Using sulfur stable isotope ratios (δ34 S) for animal geolocation: Estimating the delay mechanisms between diet ingestion and isotope incorporation in tail hair.

RATIONALE: Metabolism and diet quality play an important role in determining delay mechanisms between an animal ingesting an element and depositing the associated isotope signal in tissue. While many isotope mixing models assume instantaneous reflection of diet in an animal- tissue, this is rarely the case. Here we use data from wildebeest to measure the lag time between ingestion of 34 S and its detection in tail hair. METHODS: We use time-lagged regression analysis of δ34 S data from GPS-collared blue wildebeest from the Serengeti ecosystem in combination with δ34 S isoscape data to estimate the lag time between an animal ingesting and depositing 34 S in tail hair. RESULTS: The best fitting regression model of δ34 S in tail hair and an individual- position on the δ34 S isoscape is generated assuming an average time delay of 78 days between ingestion and detection in tail hair. This suggests that sulfur may undergo multiple metabolic transitions before being deposited in tissue. CONCLUSION: Our findings help to unravel the underlying complexities associated with sulfur metabolism and are broadly consistent with results from other species. These findings will help to inform research aiming to apply the variation of δ34 S in inert biological material for geolocation or understanding dietary changes, especially for fast moving migratory ungulates such as wildebeest.

Inferring spatially varying animal movement characteristics using a hierarchical continuous-time velocity model.

Understanding the spatial dynamics of animal movement is an essential component of maintaining ecological connectivity, conserving key habitats, and mitigating the impacts of anthropogenic disturbance. Altered movement and migratory patterns are often an early warning sign of the effects of environmental disturbance, and a precursor to population declines. Here, we present a hierarchical Bayesian framework based on Gaussian processes for analysing the spatial characteristics of animal movement. At the heart of our approach is a novel covariance kernel that links the spatially varying parameters of a continuous-time velocity model with GPS locations from multiple individuals. We demonstrate the effectiveness of our framework by first applying it to a synthetic data set and then by analysing telemetry data from the Serengeti wildebeest migration. Through application of our approach, we are able to identify the key pathways of the wildebeest migration as well as revealing the impacts of environmental features on movement behaviour.

Drivers of site fidelity in ungulates.

While the tendency to return to previously visited locations-termed 'site fidelity'-is common in animals, the cause of this behaviour is not well understood. One hypothesis is that site fidelity is shaped by an animal's environment, such that animals living in landscapes with predictable resources have stronger site fidelity. Site fidelity may also be conditional on the success of animals' recent visits to that location, and it may become stronger with age as the animal accumulates experience in their landscape. Finally, differences between species, such as the way memory shapes site attractiveness, may interact with environmental drivers to modulate the strength of site fidelity. We compared inter-year site fidelity in 669 individuals across eight ungulate species fitted with GPS collars and occupying a range of environmental conditions in North America and Africa. We used a distance-based index of site fidelity and tested hypothesized drivers of site fidelity using linear mixed effects models, while accounting for variation in annual range size. Mule deer Odocoileus hemionus and moose Alces alces exhibited relatively strong site fidelity, while wildebeest Connochaetes taurinus and barren-ground caribou Rangifer tarandus granti had relatively weak fidelity. Site fidelity was strongest in predictable landscapes where vegetative greening occurred at regular intervals over time (i.e. high temporal contingency). Species differed in their response to spatial heterogeneity in greenness (i.e. spatial constancy). Site fidelity varied seasonally in some species, but remained constant over time in others. Elk employed a 'win-stay, lose-switch' strategy, in which successful resource tracking in the springtime resulted in strong site fidelity the following spring. Site fidelity did not vary with age in any species tested. Our results provide support for the environmental hypothesis, particularly that regularity in vegetative phenology shapes the strength of site fidelity at the inter-annual scale. Large unexplained differences in site fidelity suggest that other factors, possibly species-specific differences in attraction to known sites, contribute to variation in the expression of this behaviour. Understanding drivers of variation in site fidelity across groups of organisms living in different environments provides important behavioural context for predicting how animals will respond to environmental change.

Transmission ecology of canine parvovirus in a multi-host, multi-pathogen system.

Understanding multi-host pathogen maintenance and transmission dynamics is critical for disease control. However, transmission dynamics remain enigmatic largely because they are difficult to observe directly, particularly in wildlife. Here, we investigate the transmission dynamics of canine parvovirus (CPV) using state-space modelling of 20 years of CPV serology data from domestic dogs and African lions in the Serengeti ecosystem. We show that, although vaccination reduces the probability of infection in dogs, and despite indirect enhancement of population seropositivity as a result of vaccine shedding, the vaccination coverage achieved has been insufficient to prevent CPV from becoming widespread. CPV is maintained by the dog population and has become endemic with approximately 3.5-year cycles and prevalence reaching approximately 80%. While the estimated prevalence in lions is lower, peaks of infection consistently follow those in dogs. Dogs exposed to CPV are also more likely to become infected with a second multi-host pathogen, canine distemper virus. However, vaccination can weaken this coupling, raising questions about the value of monovalent versus polyvalent vaccines against these two pathogens. Our findings highlight the need to consider both pathogen- and host-level community interactions when seeking to understand the dynamics of multi-host pathogens and their implications for conservation, disease surveillance and control programmes.

Dynamics of a morbillivirus at the domestic-wildlife interface: Canine distemper virus in domestic dogs and lions.

Morbilliviruses cause many diseases of medical and veterinary importance, and although some (e.g., measles and rinderpest) have been controlled successfully, others, such as canine distemper virus (CDV), are a growing concern. A propensity for host-switching has resulted in CDV emergence in new species, including endangered wildlife, posing challenges for controlling disease in multispecies communities. CDV is typically associated with domestic dogs, but little is known about its maintenance and transmission in species-rich areas or about the potential role of domestic dog vaccination as a means of reducing disease threats to wildlife. We address these questions by analyzing a long-term serological dataset of CDV in lions and domestic dogs from Tanzania's Serengeti ecosystem. Using a Bayesian state-space model, we show that dynamics of CDV have changed considerably over the past three decades. Initially, peaks of CDV infection in dogs preceded those in lions, suggesting that spill-over from dogs was the main driver of infection in wildlife. However, despite dog-to-lion transmission dominating cross-species transmission models, infection peaks in lions became more frequent and asynchronous from those in dogs, suggesting that other wildlife species may play a role in a potentially complex maintenance community. Widespread mass vaccination of domestic dogs reduced the probability of infection in dogs and the size of outbreaks but did not prevent transmission to or peaks of infection in lions. This study demonstrates the complexity of CDV dynamics in natural ecosystems and the value of long-term, large-scale datasets for investigating transmission patterns and evaluating disease control strategies.

Interplay of competition and facilitation in grazing succession by migrant Serengeti herbivores.

Competition, facilitation, and predation offer alternative explanations for successional patterns of migratory herbivores. However, these interactions are difficult to measure, leaving uncertainty about the mechanisms underlying body-size-dependent grazing-and even whether succession occurs at all. We used data from an 8-year camera-trap survey, GPS-collared herbivores, and fecal DNA metabarcoding to analyze the timing, arrival order, and interactions among migratory grazers in Serengeti National Park. Temporal grazing succession is characterized by a "push-pull" dynamic: Competitive grazing nudges zebra ahead of co-migrating wildebeest, whereas grass consumption by these large-bodied migrants attracts trailing, small-bodied gazelle that benefit from facilitation. "Natural experiments" involving intense wildfires and rainfall respectively disrupted and strengthened these effects. Our results highlight a balance between facilitative and competitive forces in co-regulating large-scale ungulate migrations.

Local and wide-scale livestock movement networks inform disease control strategies in East Africa.

Livestock mobility exacerbates infectious disease risks across sub-Saharan Africa, but enables critical access to grazing and water resources, and trade. Identifying locations of high livestock traffic offers opportunities for targeted control. We focus on Tanzanian agropastoral and pastoral communities that account respectively for over 75% and 15% of livestock husbandry in eastern Africa. We construct networks of livestock connectivity based on participatory mapping data on herd movements reported by village livestock keepers as well as data from trading points to understand how seasonal availability of resources, land-use and trade influence the movements of livestock. In communities that practise agropastoralism, inter- and intra-village connectivity through communal livestock resources (e.g. pasture and water) was 1.9 times higher in the dry compared to the wet season suggesting greater livestock traffic and increased contact probability. In contrast, livestock from pastoral communities were 1.6 times more connected at communal locations during the wet season when they also tended to move farther (by 3 km compared to the dry season). Trade-linked movements were twice more likely from rural to urban locations. Urban locations were central to all networks, particularly those with potentially high onward movements, for example to abattoirs, livestock holding grounds, or other markets, including beyond national boundaries. We demonstrate how livestock movement information can be used to devise strategic interventions that target critical livestock aggregation points (i.e. locations of high centrality values) and times (i.e. prior to and after the wet season in pastoral and agropastoral areas, respectively). Such targeted interventions are a cost-effective approach to limit infection without restricting livestock mobility critical to sustainable livelihoods.

Deep learning enables satellite-based monitoring of large populations of terrestrial mammals across heterogeneous landscape.

New satellite remote sensing and machine learning techniques offer untapped possibilities to monitor global biodiversity with unprecedented speed and precision. These efficiencies promise to reveal novel ecological insights at spatial scales which are germane to the management of populations and entire ecosystems. Here, we present a robust transferable deep learning pipeline to automatically locate and count large herds of migratory ungulates (wildebeest and zebra) in the Serengeti-Mara ecosystem using fine-resolution (38-50 cm) satellite imagery. The results achieve accurate detection of nearly 500,000 individuals across thousands of square kilometers and multiple habitat types, with an overall F1-score of 84.75% (Precision: 87.85%, Recall: 81.86%). This research demonstrates the capability of satellite remote sensing and machine learning techniques to automatically and accurately count very large populations of terrestrial mammals across a highly heterogeneous landscape. We also discuss the potential for satellite-derived species detections to advance basic understanding of animal behavior and ecology.

Body size and the division of niche space: food and predation differentially shape the distribution of Serengeti grazers.

1. Theory predicts that small grazers are regulated by the digestive quality of grass, while large grazers extract sufficient nutrients from low-quality forage and are regulated by its abundance instead. In addition, predation potentially affects populations of small grazers more than large grazers, because predators have difficulty capturing and handling large prey. 2. We analyse the spatial distribution of five grazer species of different body size in relation to gradients of food availability and predation risk. Specifically, we investigate how the quality of grass, the abundance of grass biomass and the associated risks of predation affect the habitat use of small, intermediate and large savanna grazers at a landscape level. 3. Resource selection functions of five mammalian grazer species surveyed over a 21-year period in Serengeti are calculated using logistic regressions. Variables included in the analyses are grass nitrogen, rainfall, topographic wetness index, woody cover, drainage lines, landscape curvature, water and human habitation. Structural equation modelling (SEM) is used to aggregate predictor variables into 'composites' representing food quality, food abundance and predation risk. Subsequently, SEM is used to investigate species' habitat use, defined as their recurrence in 5 × 5 km cells across repeated censuses. 4. The distribution of small grazers is constrained by predation and food quality, whereas the distribution of large grazers is relatively unconstrained. The distribution of the largest grazer (African buffalo) is primarily associated with forage abundance but not predation risk, while the distributions of the smallest grazers (Thomson's gazelle and Grant's gazelle) are associated with high grass quality and negatively with the risk of predation. The distributions of intermediate sized grazers (Coke's hartebeest and topi) suggest they optimize access to grass biomass of sufficient quality in relatively predator-safe areas. 5. The results illustrate how top-down (vegetation-mediated predation risk) and bottom-up factors (biomass and nutrient content of vegetation) predictably contribute to the division of niche space for herbivores that vary in body size. Furthermore, diverse grazing assemblages are composed of herbivores of many body sizes (rather than similar body sizes), because these herbivores best exploit the resources of different habitat types.

Livestock movement informs the risk of disease spread in traditional production systems in East Africa.

In Africa, livestock are important to local and national economies, but their productivity is constrained by infectious diseases. Comprehensive information on livestock movements and contacts is required to devise appropriate disease control strategies; yet, understanding contact risk in systems where herds mix extensively, and where different pathogens can be transmitted at different spatial and temporal scales, remains a major challenge. We deployed Global Positioning System collars on cattle in 52 herds in a traditional agropastoral system in western Serengeti, Tanzania, to understand fine-scale movements and between-herd contacts, and to identify locations of greatest interaction between herds. We examined contact across spatiotemporal scales relevant to different disease transmission scenarios. Daily cattle movements increased with herd size and rainfall. Generally, contact between herds was greatest away from households, during periods with low rainfall and in locations close to dipping points. We demonstrate how movements and contacts affect the risk of disease spread. For example, transmission risk is relatively sensitive to the survival time of different pathogens in the environment, and less sensitive to transmission distance, at least over the range of the spatiotemporal definitions of contacts that we explored. We identify times and locations of greatest disease transmission potential and that could be targeted through tailored control strategies.

Landscape-scale analyses suggest both nutrient and antipredator advantages to Serengeti herbivore hotspots.

Mechanistic explanations of herbivore spatial distribution have focused largely on either resource-related (bottom-up) or predation-related (top-down) factors. We studied direct and indirect influences on the spatial distributions of Serengeti herbivore hotspots, defined as temporally stable areas inhabited by mixed herds of resident grazers. Remote sensing and variation in landscape features were first used to create a map of the spatial distribution of hotspots, which was tested for accuracy against an independent data set of herbivore observations. Subsequently, we applied structural equation modeling to data on soil fertility and plant quality and quantity across a range of sites. We found that hotspots in Serengeti occur in areas that are relatively flat and located away from rivers, sites where ungulates are less susceptible to predation. Further, hotspots tend to occur in areas where hydrology and rainfall create conditions of relatively low-standing plant biomass, which, coupled with grazing, increases forage quality while decreasing predation risk. Low-standing biomass and higher leaf concentrations of N, Na, and Mg were strong direct predictors of hotspot occurrence. Soil fertility had indirect effects on hotspot occurrence by promoting leaf Na and Mg. The results indicate that landscape features contribute in direct and indirect ways to influence the spatial distribution of hotspots and that the best models incorporated both resource- and predation-related factors. Our study highlights the collective and simultaneous role of bottom-up and top-down factors in determining ungulate spatial distributions.

Author Correction: Longest terrestrial migrations and movements around the world.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Cross-boundary human impacts compromise the Serengeti-Mara ecosystem.

Protected areas provide major benefits for humans in the form of ecosystem services, but landscape degradation by human activity at their edges may compromise their ecological functioning. Using multiple lines of evidence from 40 years of research in the Serengeti-Mara ecosystem, we find that such edge degradation has effectively "squeezed" wildlife into the core protected area and has altered the ecosystem's dynamics even within this 40,000-square-kilometer ecosystem. This spatial cascade reduced resilience in the core and was mediated by the movement of grazers, which reduced grass fuel and fires, weakened the capacity of soils to sequester nutrients and carbon, and decreased the responsiveness of primary production to rainfall. Similar effects in other protected ecosystems worldwide may require rethinking of natural resource management outside protected areas.

Longest terrestrial migrations and movements around the world.

Long-distance terrestrial migrations are imperiled globally. We determined both round-trip migration distances (straight-line measurements between migratory end points) and total annual movement (sum of the distances between successive relocations over a year) for a suite of large mammals that had potential for long-distance movements to test which species displayed the longest of both. We found that caribou likely do exhibit the longest terrestrial migrations on the planet, but, over the course of a year, gray wolves move the most. Our results were consistent with the trophic-level based hypothesis that predators would move more than their prey. Herbivores in low productivity environments moved more than herbivores in more productive habitats. We also found that larger members of the same guild moved less than smaller members, supporting the 'gastro-centric' hypothesis. A better understanding of migration and movements of large mammals should aid in their conservation by helping delineate conservation area boundaries and determine priority corridors for protection to preserve connectivity. The magnitude of the migrations and movements we documented should also provide guidance on the scale of conservation efforts required and assist conservation planning across agency and even national boundaries.

From single steps to mass migration: the problem of scale in the movement ecology of the Serengeti wildebeest.

A central question in ecology is how to link processes that occur over different scales. The daily interactions of individual organisms ultimately determine community dynamics, population fluctuations and the functioning of entire ecosystems. Observations of these multiscale ecological processes are constrained by various technological, biological or logistical issues, and there are often vast discrepancies between the scale at which observation is possible and the scale of the question of interest. Animal movement is characterized by processes that act over multiple spatial and temporal scales. Second-by-second decisions accumulate to produce annual movement patterns. Individuals influence, and are influenced by, collective movement decisions, which then govern the spatial distribution of populations and the connectivity of meta-populations. While the field of movement ecology is experiencing unprecedented growth in the availability of movement data, there remain challenges in integrating observations with questions of ecological interest. In this article, we present the major challenges of addressing these issues within the context of the Serengeti wildebeest migration, a keystone ecological phenomena that crosses multiple scales of space, time and biological complexity.This article is part of the theme issue 'Collective movement ecology'.

Serengeti wildebeest migratory patterns modeled from rainfall and new vegetation growth.

We used evolutionary programming to model innate migratory pathways of wildebeest in the Serengeti Mara Ecosystem, Tanzania and Kenya. Wildebeest annually move from the southern short-grass plains of the Serengeti to the northern woodlands of the Mara. We used satellite images to create 12 average monthly and 180 10-day surfaces from 1998 to 2003 of percentage rainfall and new vegetation. The surfaces were combined in five additive and three multiplicative models, with the weightings on rainfall and new vegetation from 0% to 100%. Modeled wildebeest were first assigned random migration pathways. In simulated generations, animals best able to access rainfall and vegetation were retained, and they produced offspring with similar migratory pathways. Modeling proceeded until the best pathway was stable. In a learning phase, modeling continued with the ten-day images in the objective function. The additive model, influenced 25% by rainfall and 75% by vegetation growth, yielded the best agreement, with a multi-resolution comparison to observed densities yielding 76.8% of blocks in agreement (kappa = 0.32). Agreement was best for dry season and early wet season (kappa = 0.22-0.57), and poorest for the late wet season (0.04). The model suggests that new forage growth is a dominant correlate of wildebeest migration.

Assessing Rotation-Invariant Feature Classification for Automated Wildebeest Population Counts.

Accurate and on-demand animal population counts are the holy grail for wildlife conservation organizations throughout the world because they enable fast and responsive adaptive management policies. While the collection of image data from camera traps, satellites, and manned or unmanned aircraft has advanced significantly, the detection and identification of animals within images remains a major bottleneck since counting is primarily conducted by dedicated enumerators or citizen scientists. Recent developments in the field of computer vision suggest a potential resolution to this issue through the use of rotation-invariant object descriptors combined with machine learning algorithms. Here we implement an algorithm to detect and count wildebeest from aerial images collected in the Serengeti National Park in 2009 as part of the biennial wildebeest count. We find that the per image error rates are greater than, but comparable to, two separate human counts. For the total count, the algorithm is more accurate than both manual counts, suggesting that human counters have a tendency to systematically over or under count images. While the accuracy of the algorithm is not yet at an acceptable level for fully automatic counts, our results show this method is a promising avenue for further research and we highlight specific areas where future research should focus in order to develop fast and accurate enumeration of aerial count data. If combined with a bespoke image collection protocol, this approach may yield a fully automated wildebeest count in the near future.