Tuberculosis Epidemiology and Spatial Ecology at the Cattle-Wild Boar Interface in Northern Spain

Tuberculosis (TB) is a contagious chronic disease due to infection with Mycobacterium tuberculosis complex (MTC) bacteria. Monitoring of wildlife, especially potential reservoirs, is important for detecting changes in disease occurrence and assessing the impact of interventions. Here, we examined whether wild boar ( Sus scrofa ) may contribute to the re-emergence of TB in Asturias (10,604km 2 ), northern Spain. Although this province was declared free of TB in cattle in November 2021, MTC bacteria remain prevalent in several “hotspots,” withthe European badger ( Melesmeles ) suggested asa TB potential wild reservoir.Drawing on datafrom theSpanish National Bovine Tuberculosis Eradication Program and the Government of the Principality of Asturias covering the period 2014–2020, we analyzed the prevalence of TB in cattle and wild boar in this region. In hotspots (592km 2 ), we also investigated the ranging behavior and habitat use offvecows thatbelonged tofarmswithahistoryofTBandsixtrappedsympatric wildboar.Duringtheobservationperiod,TBprevalence was 0.14% among cattle overall and 0.13–0.41% in hotspots, which was much lower than the prevalence in wild boar, which was 3.15% overalland 5.23–5.96%in hotspots.Infected cattleandinfectedwildboar inhotspotsshared thesamestrainsof M.bovis ,andGPStracking showed spatiotemporal overlap between the species, mainly around pastures during sunrise (06:00–07:00h) and sunset (19:00–20:00h). Our results suggest that in addition to cattle and badgers, wild boar possibly help maintain TB in northern Spain, increasing the host richness that infuences TB transmission risk in the area, which should be taken into account in monitoring and eradication eforts.


Introduction
Tuberculosis (TB) is a contagious chronic disease due to infection with Mycobacterium tuberculosis complex (MTC) bacteria, principally M. bovis and M. caprae [1].Te disease is a major social, economic, and public health challenge, afecting domestic and wild animals [2].Reservoirs of the disease among wildlife can reduce the efcacy of eforts to eradicate it from cattle [3].TB in Europe afects multiple species, such that most infected animals are not bovine [4], implying a wide range of potential reservoirs.For example, the native Eurasian wild boar (Sus scrofa), whose populations are growing in Europe [5,6], can contribute substantially to TB epidemiology in southwestern Spain and France [7][8][9].On the other hand, the European badger (Meles meles) is the reservoir of TB in the British Islands [10].Te infuence of these and other wild TB reservoirs depends on numerous factors, including the prevalence of TB among cattle, characteristics and density of reservoir population, land use, as well as cattle and wildlife management practices [11][12][13].
Te abundance of potential TB reservoirs in the wild highlights the need for integrated wildlife monitoring in order to identify changes in disease occurrence and assess the impact of interventions [14][15][16].Tis is particularly true in "hotspots" where MTC bacteria have become endemic and remain prevalent in cattle [17].Such monitoring should examine the epidemiology and temporal dynamics of TB [18,19] as well as the spatial ecology of wildlife [11,20].A particularly useful tool in these eforts is the global positioning system (GPS) technology, which can monitor individuals and characterize the network of interactions within and between potential hosts to clarify how pathogens persist and spread between livestock and wildlife [21][22][23][24].
Asturias (10,604 km 2 ), in northern Spain, was declared free of TB in cattle in November 2021, Implementing Regulation (UE) 2021/1911 [25], but several hotspots persist, and the region appears to contain potential disease reservoirs, i.e., badger [26].Disease transmission between badgers and cattle has been documented and supported by, for example, GPS studies showing badgers' presence in cattle paddocks [20,26].In this way, Asturias is an excellent example of a region where a comprehensive understanding of TB persistence and transmission in multiple wild hosts is essential to control the disease.
To clarify to what extent the interaction between wild boar and cattle helps maintain TB in this area, we analyzed disease prevalence in these species in hotspots in Asturias during the period 2014-2020.We also used GPS collars to track the ranging behavior and habitat use of wild boar in relation to cattle.Our results may help clarify how wild hosts can maintain TB in endemic hotspots within an area declared free of the disease in cattle.

Ethical Statement.
All methods were carried out in accordance with relevant guidelines and regulations [27].All experimental protocols were approved by ethical committees from the Government of Principality of Asturias, license reference number: PROAE 47/2018.

Study Area.
Te study was carried out in three TB hotspots in Piloña (A) and Caso (B, C) in Asturias (Figure 1), which are located in northwestern Spain (43 °21′ N; 5 °57′ W).Tis region is characterized by an Atlantic climate, abundant precipitations (>1000 mm/year), soft temperatures, and small thermal amplitude [28].Te land varies from 0 m to 2650 m above sea level ("Torrecerredo" peak) and is covered by 1.8% of cultures, 29.4% of pastures and 57.5% of forested land, approximately [29].Here, large-scale fencing for cattle does not exist, and wild boar, whose abundance has increased in the last years, moves freely over the territory [30].

Tuberculosis Prevalence in Cattle and Wild Boar.
Cattle TB prevalence data from 2014 to 2020 for Asturias and Caso and Piloña (see Figure 1) were obtained from the 2022 National Bovine Tuberculosis Eradication Program [25], whilst wild boar TB prevalence data were obtained from the Government of the Principality of Asturias.Data were obtained by the ofcial single intradermal tuberculin test and culture in cattle and by culture in wild boar (n � 1341).
Diferences in prevalence between species were assessed using Mann-Whitney U tests.Te statistical test was carried out using SPSS Statistics 25 (IBM, New York, USA), and the signifcance level was set at p < 0.05.

Animal Trapping and
Monitoring.Diferent cattle and wild boar were monitored in relation to their ranging behavior in Caso and Piloña (approximately an area of 592 km 2 ) (Figure 1).

2.4.1.
Cattle.Five adult cows (C1, C2, C3, C4, and C5) were monitored in the three hotspots (see Figure 1).We randomly selected cows that belonged to farms with a history of TB.Tey were tracked using Digitanimal GPS radio-collars (Digitanimal ® , Madrid, Spain), programmed to acquire one location per 30 min, including time, date, geographic coordinates, and temperature (Table 1).
Tey were monitored in diferent years and months by using Microsensory GPS radio-collars (Microsensory System, Córdoba, Spain) and Digitanimal GPS radio-collars, programmed to provide the location of the animal in a determined frequency, time, date, geographic coordinates, and temperature (Table 1).

2
Transboundary and Emerging Diseases GPS radio-collars in cattle were used for further studies, and they had a shorter programmed acquisition time and a larger monitoring period; consequently, the number of cattle locations was higher than that of wild boar.

Data Analyses.
For the 11 monitored animals (cattle and wild boar), activity patterns, home ranges (HRs), and habitat selection patterns were completed.Spatial overlap, environmental overlap, and temporal overlap between species were analyzed to describe the potential of species interactions.

Activity Pattern.
To calculate the animal's activity pattern, a straight line was obtained between consecutive GPS locations separated by intervals of 2 hours (h), as locations of wild boar were programmed to acquire information every 2 h (W4 and W5 were not considered due to the lack of locations, see Table 1).Tis distance was then divided by the time elapsed between them (km/h) and was used to infer activity patterns.

Temporal Overlap.
Overlap analysis for cattle and wild boar activity patterns acquired was undertaken using the "overlap" R package [35].Overlap coefcients (Δ) went from 0 (no overlap) to 1 (full overlap), and the 95% confdence intervals (CI) were obtained through 1000 bootstrap samples.As all samples were >50 records, the coefcient Δ4 was considered [36].

Home Range.
Annual and seasonal HRs were estimated using the fxed-kernel function in the "adehabitat" R Table 1: Information provided by the GPS of the monitored animals.Identifcation given to each animal, months and years in which the GPS provided information, acquisition programmed time (frequency of locations), number of days with data, and number of locations given.Note that the number of cattle locations is higher than the number of wild boar locations due to shorter programmed acquisition time and larger monitoring period.package, R version 3.6.1 in [35] program [37].Kernel 95% was used to indicate the home range (HR), and kernel 50% the core range (CR) [38].Te least-squarescross-validation method failed to converge in animals with large sample sizes; therefore, kernels were estimated using the reference bandwidth method [39,40].According to previous studies and to estimate HRs [20,40], the minimum number of relocations per individual was established at 25. Diferences in HR and CR among seasons and species were assessed using Kruskal-Wallis H tests and Mann-Whitney U tests, respectively.

Spatial Overlap.
Home range and CR were used to estimate spatial overlap between wild boar and cattle within the study area.Spatial overlap was calculated using the overlap function in "rgeos" R package [35].Tis gave information on the area intersected (HR and CR) between cattle and wild boar relative to cattle when the area was divided by the HR or CR of the wild boar and vice versa when it was divided by the HR or CR of the cattle [41].

Environmental Overlap.
Land uses of Caso and Piloña were obtained from a combination of the Corine Land Cover [42] and the National Center for Geographic Information [43], with scale cartography of 1:250,000 and considering the following land uses: cultures, pastures, woodland, shrubland, water, no vegetation areas, and urban areas, as seen in previous studies [20,44,45], in view of biological relevance.In this study, urban areas were villages with low population density (<27 inhab/km 2 ), with economical activities mainly based on agricultural and livestock production.
GPS locations were bufered to assess interactions between cattle and wild boar in diferent land uses.Tese considered GPS positional error and missing locations to avoid misclassifcation of habitat use, possibly given by landscape heterogeneity and lack of GPS locations [46].Te proportional cover of each land use was calculated within each bufer.Latent selection diference functions (LSDs) were estimated using logistic regression and the "rms" R package [35], as described in Barasona et al. [11].In these analyses, locations of cattle were coded as 1 and locations of wild boar as 0, so cattle resource selection or avoidance relative to wild boar could be evaluated.Variables with signifcant positive coefcients showed preferred land uses by cattle relative to wild boar, and those with signifcant negative coefcients indicated avoided land uses.However, variables with no signifcant coefcients indicated areas with no diference of use and, therefore, with a high potential for interaction.
Te Huber-White sandwich estimator was used to estimate standard errors, grouping data by the individual, which considered an unbalanced sampling design and nonindependence of observations belonging to the same individual [47].Te best model was considered by means of a forwards-backwards stepwise procedure based on Akaike information criteria (AIC) [48].2).Signifcant diferences were observed between species (p � 0.024).
Both in Caso and Piloña, the same M. bovis isolates were characterized in cattle and wild boar from the same region.Te identifed 4 spoligotypes and VNTR profles are shown in Table 2.

Activity Pattern and Temporal Overlap.
To compare activity patterns of cattle and wild boar, and due to the absence of information on wild boar GPS monitoring in autumn, winter, and spring, activity patterns were only obtained for summer.Cattle manifested one peak at 06:00 h and a similar movement until 20:00 h, when activity started to decrease.Wild boar, on the other side, exhibited two diferent peaks in their activity: frst at around 05:00 h and second at around 21:00 h, coinciding with sunrise and sunset approximately (Figure 3).However, considering both species, two diferent periods could be established: (1) from 08: 00 h to 20:00 h and (2) from 20:00 h to 08:00 h.Te 4 Transboundary and Emerging Diseases coefcient of overlap (Δ4) resulted in 0.63 (confdence intervals 0.54-0.73)(Figure 3).4).Signifcant differences were found in cattle HR and CR among seasons (Kruskal-Wallis χ 2 = 9.817, p � 0.020; χ 2 = 9.307, p � 0.025, respectively).

Spatial
Overlap.Spatial overlaps between cattle and wild boar in HR and CR difered among areas during summer, observing the highest overlap in the hotspot of Piloña (A), followed by hotspot "B" and hotspot "C" in Caso (Figure 5).Overall, >32% of the cattle HR overlapped wild boar HR, whereas around 17% of the wild boar HR overlapped cattle HR.When comparing summer wild boar HR to the annual HR sizes of cattle, overlaps increased in most of them, either HR or CR.In this case, overall, >40% of the cattle HR overlapped with wild boar HR, whereas 28% of the wild boar HR overlapped with cattle HR (Table 3).

Environmental Overlap.
Models selected woodland and pastures as areas that better segregated cattle and wild boar (coefcients resulted signifcant or marginally signifcant), whereas cultures, shrubland, and urban areas as land use which worst segregated both species (not signifcant coefcients).Cattle showed avoidance of areas with a higher proportion of woodland, relative to wild boar, and preferred areas with a higher proportion of pastures, also relative to wild boar (Table 4).
Te model selected, which better explained the habitat selection of cattle relative to wild boar, chose as infuential variables woodland, urban areas, pastures and shrubland.Te resulting model was signifcant (p � < 0.0001) and with an R 2 = 0.671.

Discussion
Our 7-year analysis of Asturias indicates that the prevalence of TB among wild boar, particularly in hotspots, has gradually increased, and we have provided evidence that the disease moves between wild boar and cattle.Tese fndings establish wild boar, together with badger [26], as wild reservoirs of the disease in hotspots, which helps clarify how TB remains a threat to cattle in these areas.
Te same M. bovis strains and VNTR profles were identifed from infected wild boar and infected cattle in our study.Given that TB prevalence among cattle in Asturias increased from 2014 to 2015 and then increased among wild boar from 2016-2018, we speculate that TB might have travelled from cattle to wild boar during the study period.Indeed, an initial increase in disease prevalence among cattle followed by an increase among wild boar was observed in the hotspots of Caso and Piloña.Te slight increase in TB prevalence among cattle from 2018 (0.05%) to 2019-2020 (0.09%), in turn, suggests that either other infected cattle or wildlife (i.e., wild boar and badger) might be the source of infection in TB-free farms.Wild boar in Europe have been Transboundary and Emerging Diseases shown to transmit TB efciently to cattle [49].In addition, the population density and distribution of wild boar have increased in northern Spain since the 1990s [30,50].Te wild boar may continue to grow as a TB threat to cattle in this area, given their ability to thrive in a wide range of habitats [51].
Te cattle in our study were active mainly in the middle of the day, whereas wild boar were active mainly in the early morning and late afternoon.Nevertheless, we identifed two periods when both species were highly active: 06:00-07:00 h and 19:00-20:00 h.At these times, the two species likely came into contact, given the spatial overlap in their distribution.Tus, restricting cattle movements at sunrise and sunset may help protect them from infected wild boar.Such measures may be particularly important in the hotspot of Piloña, where the two species showed the greatest spatial overlap and the highest TB prevalence among wild boar (2018).
Te HR and CR of cattle in our study were signifcantly larger in autumn and summer than in spring and winter, consistent with the production systems common in Asturias [52].In such systems, animals are kept in communal pastures and open meadows in spring and summer, with short periods of stabling during winter.In summer, the HR and CR of wild boar were even larger than those of cattle.As cattle and wild boar share habitat use with no or little restrictions in these traditional communal pastures, diferent herds, as well as wildlife species, can circulate among the same pastures, increasing the risk of disease transmission.Our analysis, therefore, strengthens the case for biosecurity measures in this region [24,53].
We analyzed the HR of wild boar only during summer, neglecting the hunting season from September to February.During the summer, this HR substantially overlapped with the HR of cattle, implying that the two HRs overlapped to at least some extent during other seasons, including when cattle were in paddocks.Indeed, the HR of wild boar in Asturias remained large throughout the year, though it increased signifcantly during the hunting season [54].We speculate that wild boar in Asturias may reach even larger numbers of cattle in the autumn than in the summer.On the other hand, hunting can substantially reduce the number of wild boar [30,55,56].In fact, an efcient hunting season may explain the lower TB prevalence among wild boar in 2018-2019, although infuence from diferences in feld sampling and diagnostic testing cannot be excluded [57].Future work should consider the efects of hunting season on the TB risk that wild boar poses to cattle.In addition to hunting by humans, predation by gray wolves (Canis lupus) can reduce wild boar populations, especially the numbers of TB sickly animals, which are more likely to shed pathogen into the environment, and the numbers of piglets, which are more susceptible to infection.Such predation of wildlife reservoirs has been proposed as a major form of natural infection control [58,59].
While wildlife reservoirs are a major source of infection for cattle [60,61], other factors also contribute to infection risk, including the size of the cattle herd, the number of incoming animals in recent years, pasture lease agreements and transhumance to areas with high TB prevalence [62][63][64].Future work should explore the full range of factors driving TB prevalence among cattle in Asturias, which may help clarify disease maintenance in other TB-free areas.Such work should carefully consider additional factors that may infuence the risk of disease transmission directly or indirectly.For example, although our study refects that cattle prefer pastures and wild boar prefer woodlands because they provide shelter [45], the cattle dung on pastures favors the presence of earthworms [65], which are known to attract wild boar [66,67] and also badgers [20], potentially increasing interactions and infection transmission.At the same time, wild boar is expanding into more humanized landscapes, usually when their nutritional needs are not met [68].Aside from landscape type and land use, environmental contamination may be important for maintaining TB in an area, given that M. bovis can persist in favorable environments for long periods [69][70][71].

Transboundary and Emerging Diseases
Our work underscores that TB in some regions of Europe is a truly multi-host disease that involves cattle as well as domestic and wild nonbovine species [4].Host richness is an important factor infuencing the transmission risk of infectious diseases, including TB [72].TB in northern Spain is known to be maintained jointly by cattle and badger [26], to which the present work adds wild boar.Te dynamics of infection among all these epidemiologically relevant hosts should be considered if TB control programs are to be efective [4,72].
Our fndings in this study, however, should be interpreted with caution as the results in activity pattern, and temporal, spatial, and environmental overlap may be biased due to insufcient animals, lack of locations and short periods of monitoring, thus representing a conservative fgure of the epidemiology and spatial ecology in the area.

Conclusions
Our work shows evidence of spatiotemporal overlap between cattle and wild boar in areas with high TB prevalence, mainly around pastures during sunrise and sunset.We also indicate a possible interspecies TB transmission, likely from cattle to wild boar, although TB prevalence trends suggest that TB-free cattle could also be infected by wild boar.Terefore, in addition to cattle and badger, wild boar can help maintain TB in northern Spain, increasing the host richness that infuences TB transmission risk in the area.

Figure 2 :
Figure 2: Cattle and wild boar tuberculosis (TB) prevalence trends.Information is given for the region of Asturias and the regions of Caso and Piloña (where the hotspots remain) during 2014-2020.

Figure 3 :Figure 5 :
Figure 3: Cattle and wild boar activity patterns and overlap among them.Animals were monitored in the three hotspots, and activities were measured for summer.Overlap (gray shade) between cattle's activity pattern (blue dotted line) and wild boar's activity pattern (black line) among the diferent hours of the day is indicated.

Figure 4 :
Figure 4: Cattle and wild boar home range (HR) and core range (CR).(a) HR (red) and CR (gray) of cattle in summer, autumn, winter, and spring.(b) HR of cattle (red) and wild boar (dark blue) and CR (gray) for both species in summer.Error bars indicate standard deviations.
Cattle and Wild Boar.Te prevalence of TB in cattle in Asturias decreased between

Table 2 :
Mycobacterium bovis isolates characterized for cattle and wild boar in the tuberculosis hotspots from Caso and Piloña.Spoligotype (SB) number, VNTR profle, number (n) of cattle and wild boar, and total of animals characterized are indicated.