Understanding the spatial patterns and drivers of animal movement is a crucial first step to controlling disease spread4. Our study provides novel information about where, how and when cattle move in a region beset by endemic pathogens2,39,40. Because contacts occur heterogeneously through time and space, interventions targeting areas and times of high contact risk could effectively break the chain of transmission across wide areas. We found that cattle herds had the highest probability of contact at dipping sites, far from their bomas, in small herds and during periods of low rainfall, indicating that transmission of all pathogens may be particularly elevated under these conditions (Figs. 5, 6). Nonetheless, cattle spent most of their time in other areas (i.e. near bomas or in grazing areas) where the direction and magnitude of effect of spatiotemporal scale on contact rates varies. This suggests that interventions for different pathogens in these systems will likely require a consideration of scale of transmission and be tailored to particular pathogens. Overall, our study provides a framework for risk-based livestock disease control approaches for the most dominant management systems in sub-Saharan Africa.
Daily movement patterns of cattle in pastoral and agropastoral settings in sub-Saharan Africa largely reflect the distribution of shared resources, which determines the distance animals move each day and the probability of contacting each other. Our results are similar to those reported in other regions of Africa, suggesting broadly comparable patterns of daily displacement. For instance, cattle in our agropastoral study area travel to grazing, watering and dipping locations that are ~ 4 km from their bomas and primarily during daylight hours (Fig. 2). Similarly, in Kenya, cattle in the pastoral Mara and Ol Pajeta regions move less than 6 km from their bomas and movements peak around 12:00–14:00 h each day9,41. Despite the predominance of short-distance daily movements, we observed occasional long-distance movements (i.e. up to 12 km), particularly by larger herds. Transhumant cattle in Cameroon also moved up to 23 km/day for short periods, while relocating to seasonal grazing areas on the edge of the Sahel, though in most observations (86%) they moved less than 5 km/day8. Although we observed no contacts among cattle from bomas > 17 km apart (Supplementary Fig. S5), regardless of how contact was defined, infrequent long-distance movements by large herds may provide a conduit for disease transmission between villages42. Indeed, larger herds actually had a lower relative probability of contact across spatiotemporal scales (Fig. 5), which may reflect the fact that large herds were more likely to move to areas away from other collared cattle, either because they were moving outside the study area, or because they had exclusive use of particular areas, whereas smaller herds that were mostly moved around bomas mixed more frequently. While interventions (e.g. vaccination or quarantine) targeting small herds would address local disease events, particularly within villages, halting larger-scale transmission requires an understanding of livestock pathways enabling inter-village connectivity and strategies tailored to herds driving these processes.
A key difference between the movement of cattle in agropastoral and pastoral systems lies in the seasonal variation of daily movement. In our study, agropastoralists move their herds farther in the wet compared to the dry season, while the opposite has been reported for pastoralists8,9,41. During the wet season, agropastoralists cultivate crops near their homesteads, which increases competition for space and displaces cattle to reserved grazing areas far from cultivated land11. During the dry season, particularly in the early period, cattle graze harvested fields around the homestead and tend to move short distances each day. In our study, although individual herds travelled more (marginally) in the wet compared to the dry season, there were more contacts following low rainfall periods when resources were typically scarce (Fig. 5). Similarly, a previous study has shown that more villages were connected at shared resource areas during dry spells, which resulted in higher contacts11. This suggests a higher disease risk in the dry compared to wet seasons in agropastoral management systems.
Translating movements into contact between individuals is challenging because the definition of a “contact” depends on the distance at which pathogens can travel in space, and the time period that pathogens survive, or mature to an infectious state, in the environment. Most studies that attempt to measure contact, however, focus only on a single scale. Here, we show that pairwise contact rates between cattle herds generally increase with broader spatiotemporal definitions of contact. Yet, there was no difference at spatial scales between 50 m, 100 m and 200 m for a temporal scale of one hour, suggesting these scales are functionally equivalent definitions of contact. Thus, we define “close contact” as proximity of livestock herds within 200 m in any given hour, which would be applicable to multiple disease systems and vital for understanding infectious disease spread in traditionally managed herds. However, given that herds tracked in our study ranged in size from 30 to 500 cattle, for households with herds of < 30 cattle, a meaningful spatial scale of close contact would likely be less than 200 m. Furthermore, the relative effect of temporal scale on the contact rate was not the same at all spatial scales (Fig. 4). For instance, there was a large difference in the number of contacts between 1 h, 1 day and 1 week at all spatial scales greater than 200 m. Thus, the temporal window that defines a contact has a greater influence on contact frequency than the spatial component, at least across the range of values in our study. Again, this is an important consideration when designing interventions to break transmission for specific pathogens.
Understanding how contact varies as a function of spatial and temporal scales enables us to hypothesise how transmission risk may vary as a function of a pathogen’s mode of transmission and its stability in the environment. For example, a coarse-scale definition of contact is relevant for understanding the transmission risk of pathogens that are stable in the environment. In contrast, a fine-scale definition applies to pathogens that, for optimal transmission, require hosts to be relatively close to each other in both space and time. Pathogens with multiple transmission modes, which is the most likely situation, might have risks associated with both coarse and fine-scale contacts.
Estimating the expected total number of contacts between livestock in an area illustrates the potential for disease transmission through a region. For example, consider an area with a 10 km radius (i.e. 314 km2)—in our study area this would contain about 4 villages, each having about 250 livestock-keeping households. For short-lived pathogens requiring close contact for transmission, a meaningful contact could be defined as being within 200 m within the same hour. Using the parameters from our results, the contact rate between collared cattle was observed to be 0.015 h−1 (Supplementary Fig. S6), suggesting that across all animals in the area, given the density of cattle herds, we would expect at least 1 contact every 4.76 h. However, for pathogens that spread by airborne transmission, a meaningful contact could be defined as being within 500 m within the same hour. In this case, using our observed contact rate of 0.059 h−1 between collared cattle (Supplementary Fig. S6), we estimate a contact occurring every 40 min across all cattle in the area. This illustration provides a unique perspective about the relative speed at which different pathogens could spread through an area. The next question is, given the movement patterns of livestock, where do these contacts occur and are there specific areas in the landscape that could be targeted?
Identifying where in a given landscape inter-herd contacts are most likely to occur is challenging in observational studies as inferences will be strongly affected by sampling design (i.e. which individuals have GPS collars and how many collars are deployed). We overcome this issue by estimating a relative contact probability that accounts for both the locations where contacts occur and where they could have occurred. This conditional approach shows that locations with high contact probability tend to expand as the spatiotemporal scale of contact increases, but this pattern is not consistent across all resources (Fig. 5). For instance, contact probability was high when herds were close to shared resources such as pasture and water, but the effect of distance to water on the probability of contact switched from negative to positive as the temporal and spatial scale increased from fine to coarse resolution (Fig. 5). This suggests that distance to water may pose different levels of risk depending on the pathogen’s mode (e.g. direct versus indirect) transmission.
Identifying areas of high congregation and potential contact offers opportunities for preventative interventions, especially in settings where large-scale disease mitigation efforts, for instance mass livestock vaccination, are limited, such as our study context. Sharing of communal resources is considered a major driver of infectious disease transmission in traditionally managed livestock husbandry systems42,43,44. In our study, areas of increased contact included dips and water points. Livestock tend to congregate infrequently at dips, twice a month in most cases11. However, these gatherings are exceptionally large and intense because the dipping schedule is restricted to a few events per month. This results in a significantly higher probability of contacts across all spatiotemporal scales and suggests that these areas pose very large relative transmission risks for most types of pathogens. An initial approach to reduce high livestock congregations may be to keep dips open for more days per month and to introduce a strict rota system by household. The risk of transmission around waterholes is offset by precipitation; under high rainfall conditions when access to water is unrestricted, there are few contacts between herds and transmission risk is reduced. Repeated contacts for short periods of time also occurred at watering points (i.e. for minutes to an hour, once per day generally between 13:00 and 15:00 h), which would favour pathogens with short survival times and close-contact transmission. Fewer encounters around watering points for long contact duration suggest lower risks for pathogens (such as B. anthracis) with prolonged environmental survival, especially when precipitation levels are high. In contrast, herders access multiple grazing areas each day and tend to avoid other herds to minimise mixing and theft11. Therefore, sharing pasture results in more herds but fewer close-encounter contacts. Sharing grazing areas at distinct nonoverlapping times would thus favour the transmission of pathogens with long survival in the environment or that can be transmitted from afar. Thus, for volatile pathogens, transmission risks are likely higher around shared dips or watering points. Reducing opportunities for contacts at these locations would lower exposure risks and the number of herd infections Although contact patterns can be used to infer transmission and infection risks9,23, disease spread is influenced by complex interactions between biological and ecological factors relevant to each pathogen, only some of which are outlined here. For example, for diseases that require an arthropod vector (e,g, African trypanosomiasis transmitted by tsetse flies) or an intermediate host (e.g. echinococcosis or other parasitic diseases requiring complex transmission cycles involving intermediate and definitive hosts), pathogen transmission can occur independently of direct contact amongst individuals of the species of primary concern. Similarly, water- or food-borne infections do not require close contact for transmission. Environmental and climatic drivers, such as exceptionally high and low rainfall or soil characteristics, critically influence the ecology of pathogens with an environmental phase, e.g. B. anthracis. Fomites also contribute to within and between herd transmission of FMD and occasionally, wind-borne particles of FMD virus can facilitate the spread of the disease over long distances. Consequently, our findings should be interpreted with caution in the knowledge that contacts were defined broadly, and further sources of heterogeneity are likely. In addition, our study focused on inter-herd contacts, but intra-herd dynamics may play an important role in the speed of disease spread. Furthermore, the mobility of infected animals may be impacted in ways that change contact risks, but this heterogeneity was not captured in our dataset. Ultimately, there is a need for further studies that integrates infection and transmission surveillance with herd mobility information to establish the extent to which contact risk predicts infection of different pathogens.
A fruitful area for future research would be to compare contact rates across a broad spectrum of production systems, for example from pastoral to peri-urban smallholder systems. The transition of many communities towards urbanization could change the patterns of contact, particularly because peri-urban livestock owners have alternate sources of income and often more wealth than rural subsistence farmers. Furthermore, many sub-Saharan governments are encouraging the expansion of mechanised agriculture and agro-industry (high production dairies, supplemental feeding with hay and silage, and large abattoirs to supply the export market). These cultivated sites will overlap with traditional livestock keeping areas and could dramatically change livestock movements and disease dynamics, which will require tailored mitigation strategies. Another important area for future work would be to test how the spatial pattern and availability of critical resources could reduce mixing and disease spread while improving herd management. For instance, a simulation experiment in which the number of contacts is estimated based on the fine scale movements of multiple herds in silico would allow researchers to explore the effects of adding, removing or changing the availability of resources such as dips, water holes or additional pasture at different locations in the landscape.
Finally, linking data on fine-scale movements of livestock with epidemiological data generated in near real time during outbreak investigations would help validate our models. It would also provide valuable information on locations and times that might drive transmission risks and that could therefore be targeted through tailored interventions. Specifically, for infectious diseases that transmit over limited distances and require close/direct contact with infectious animals or materials (e.g. FMD), combining livestock movement data and epidemiological modelling of outbreak information would enable us to identify central transmission nodes (e.g. shared dips and watering locations) and times of highest congregation (e.g. periods of low rainfall), and to evaluate mitigation scenarios through movement restrictions or vaccination at these key points. In areas where large-scale vaccination programmes are impractical, this information could shape the development of locally-acceptable interventions (e.g. use of acaricide hand spray and watering livestock using water troughs) that could reduce transmission at these locations and times, hence help to flatten the epidemic curve. This type of approach would also be valuable for informing and prioritising resources for vaccination. For example, previous work on FMD in these settings 2 suggests that dominant viral serotypes move slowly, one at a time, in waves across the landscape, and that this pattern may be consistent across East Africa. Spatial models of disease spread informed by timely detection of outbreaks at the serotypic level would enable us to target vaccination ahead of the wave of infection using monovalent vaccines which are more widely available than polyvalent formulations.
Source: Ecology - nature.com
