**4.1 Characterization of published articles**

Most epidemiological studies on wild boar or red deer are cross-sectional, allowing for the estimation of prevalence rates and simultaneously the identification of risk or protective factors. A few of the earliest studies were surveys; also classified as such were some molecular epidemiology articles that allow calculating prevalence rates. As knowledge of bTB on other species is more recent, a larger proportion of these studies are case reports and surveys. A comparatively large number of studies address molecular epidemiology.

Notably absent from the literature are case-control studies, which could shed light on the importance of specific determinants of disease, such as fencing and provision of feed and water. The same should be mentioned for experimental studies, were exposure to a certain determinant of disease is manipulated and the effect on disease occurrence is then measured. This design could be of great help to ascertain the role of each species in the persistence of bTB, trough manipulation of host density. The same can be said for epidemiological modelling, which could provide a theoretical framework for understanding bTB persistence in Iberian Peninsula and test the effect of different control measures (Thrushfield *et al*., 1995) and also to identify key data on host populations and wildlife tuberculosis that is missing or that is not feasible or up to date.

Most articles resort to targeted surveillance of hunted or culled animals, which allows prevalence estimation. Culling is expected to be less sex and age-biased than recreational hunting, which focuses on specific age (adults) and sex (males) classes. The hunting method used for harvesting the animals (drive hunts) is less selective than trophy hunting, allowing access also to females and juvenile/subadult animals (Fernández-Llario & Mateos-Quesada, 2003; Martínez *et al*., 2005). Hunted animals are usually considered a representative sample of the population for health monitoring, at least for non neurological or debilitating diseases (Conner *et al*., 2000). Nevertheless it should be kept in mind that some sampling biases can be present (Wilson *et al*., 2001).

characterization of populations maintaining high bTB prevalence rates despite long-term lack of contacts with cattle. Fallow deer, Barbary sheep and badger are also discussed as possible maintenance hosts, while all other reported hosts are considered spillover. Wild boar, red and fallow deer have been suggested as possible reservoirs of infection for

The most commonly identified causative agent of bTB in Iberian Peninsula has been *M. bovis*, although a small proportion (0,05, n=829) of *Mycobacterium caprae* was reported in 6/15 studies. *M. caprae* is much more frequent among isolates from wild boar (0,08, n=502) than from red deer (0,01, n=327). *Mycobacterium avium*-complex mycobacteria and other mycobacteria have also been isolated from wild hosts, but they fall out of the scope of the present review. Molecular epidemiology studies rely mostly on spoligotyping (14/15),

Most epidemiological studies on wild boar or red deer are cross-sectional, allowing for the estimation of prevalence rates and simultaneously the identification of risk or protective factors. A few of the earliest studies were surveys; also classified as such were some molecular epidemiology articles that allow calculating prevalence rates. As knowledge of bTB on other species is more recent, a larger proportion of these studies are case reports and surveys. A comparatively large number of studies address molecular

Notably absent from the literature are case-control studies, which could shed light on the importance of specific determinants of disease, such as fencing and provision of feed and water. The same should be mentioned for experimental studies, were exposure to a certain determinant of disease is manipulated and the effect on disease occurrence is then measured. This design could be of great help to ascertain the role of each species in the persistence of bTB, trough manipulation of host density. The same can be said for epidemiological modelling, which could provide a theoretical framework for understanding bTB persistence in Iberian Peninsula and test the effect of different control measures (Thrushfield *et al*., 1995) and also to identify key data on host populations and wildlife

Most articles resort to targeted surveillance of hunted or culled animals, which allows prevalence estimation. Culling is expected to be less sex and age-biased than recreational hunting, which focuses on specific age (adults) and sex (males) classes. The hunting method used for harvesting the animals (drive hunts) is less selective than trophy hunting, allowing access also to females and juvenile/subadult animals (Fernández-Llario & Mateos-Quesada, 2003; Martínez *et al*., 2005). Hunted animals are usually considered a representative sample of the population for health monitoring, at least for non neurological or debilitating diseases (Conner *et al*., 2000). Nevertheless it should be kept in mind that some sampling biases can

livestock.

**4. Discussion** 

epidemiology.

be present (Wilson *et al*., 2001).

**3.6 Molecular epidemiology** 

usually coupled with MIRU-VNTR typing (9/15) (Table 6).

tuberculosis that is missing or that is not feasible or up to date.

**4.1 Characterization of published articles** 


Table 5. Bovine tuberculosis host species described in the Iberian Peninsula. For references and meta prevalence rate calculations for wild boar, red and fallow deer only targeteddesign studies using bacteriological culture as diagnostic test on all animals samples are used. Meta prevalence in carnivores is exclusively based on passive-design studies.

Wildlife Tuberculosis: A Systematic Review of the Epidemiology in Iberian Peninsula 285

spoligotypes

spoligotypes 13 genotypes combined

spoligotypes

spoligotypes

Table 6. Molecular biology studies included in the analysis. SP: spoligotyping, MV: MIRU-VNTR mycobacterial interspersed repetitive units-variable number of tandem repeats.

On the other hand, studies of wild ungulates relying on routine meat inspection for detection of macroscopic tuberculosis-like lesions, do not allow for a reliable estimation of prevalence, which is underestimated in this situation (de Mendonza *et al*., 2006). Nevertheless this type of design allows increasing sample size, which makes them suited for long-term surveillance rather than detailed epidemiological studies (de Mendonza *et al*., 2006) and were mostly used in the first surveys and cross-sectional studies after bTB was detected in wildlife in Iberian Peninsula. The investigations on carnivore species, most of which are not hunted, tend to rely on passive surveillance schemes based on haphazardly found carcasses. This sampling design does not allow to estimate prevalence rates due to extensive sampling bias (e.g. Taylor *et al*., 2002). Targeted sampling in these species has been attempted using serological tests but results should be interpreted with caution since these

The number of animals studied is usually adequate to determine prevalence rates with relatively small confidence intervals, at least in the easily available hunted species. The same cannot be said for most studies on protected carnivore species, where the collection of

Bacteriological culture is the reference test for diagnosing bTB although it is expensive and time-consuming (de Lisle *et al*., 2002). As the financial resources needed to perform bacteriological culture on a large number of samples are scarcely available, most surveys use

2008-2009 27

2006-2007 9

1992-2009 15

SP 2008-2009 8

2 spoligotypes red fox (none

21 spoligotypes red deer (11

15 spoligotypes wild boar (5

6 spoligotypes exclusive of wildlife vs 23 spoligotypes exclusive of

8 genotypes red deer (2 exclusive) 6 genotypes fallow deer (none

8 spoligotypes red deer (4 exclusive) 4 spoligotypes wild boar (none

5 genotypes wild boar (none

4 spoligotypes wild boar (none

1 spoligotype red (none exclusive) 1 spoligotype red fox (none exclusive) 12 spoligotypes goat (6 exclusive) 9 spoligotypes cattle (2 exclusive) 2 spoligotypes sheep, pig (none)

4 regions South-Central Portugal

1 area SW Spain

1 area Central Portugal

Spain

1 spoligotype badger (none exclusive) 239 spoligotypes cattle (207 exclusive) 3 spoligotypes goat (1 exclusive) 2 spoligotypes pig (none exclusive) 3 spoligotypes cat (1 exclusive) 1 spoligotype dog (none exclusive)

exclusive)

exclusive)

exclusive)

exclusive)

exclusive)

exclusive)

exclusive)

domestic species

(5585 cattle, 33 goat, 7 pig, 3 cat, 1 dog)

74 red deer 36 wild boar

24 red deer 21 fallow deer 62 wild boar

27 red deer 21 wild boar

14 wild boar 1 red deer 1 red fox (542 goat, 229 cattle, 2 sheep, 2 pig)

SP MV

SP MV

SP MV

techniques have not yet been validated in these species.

biological samples from a large number of animals is inherently difficult.

Cunha *et al*. (2012)

Gortázar *et al*. (2011)

Pinto *et al*. (2011)

Rodriguez *et al*. (2011)


n

spoligotypes (2 clusters)

spoligotypes 43 combined (14 clusters, 21 unique profiles)

spoligotypes

spoligotypes 19 combined

spoligotypes 131 combined (28 clusters, 76 unique profiles)

10 unique profiles)

spoligotypes

spoligotypes

genotypes

spoligotypes

spoligotypes

Host clustering Study areas

1 area W Spain

7 areas SW Spain

24 areas SW Spain

1 area W Spain

1 area W Spain

Portugal

1 area SW Spain

Portugal

3 areas Portugal

Spain

Sheep/goat isolates clustered apart

from other species

4 Iberian pig-only clusters 7 wild boar-only clusters 2 common clusters (14 genotypes)

17 genotypes in wild boar (4

6 fallow deer (1 exclusive) 10 cattle (3 exclusive)

10 spoligotypes wild boar (5

8 genotypes red deer (none exclusive)

6 spoligotypes red deer (1 exclusive)

11 spoligotypes red deer (2 exclusive) 5 spoligotypes wild boar (none

27 spoligotypes cattle (15 exclusive)

2 spoligotypes red & fallow deer &

11 spoligotypes cattle (8 exclusive)

12 genotypes red deer (8 exclusive) 4 genotypes wild boar (1 exclusive) 78 genotypes cattle (71 exclusive)

26 spoligotypes wild boar (6

1 spoligotype chamois (none

22 spoligotypes red deer (2 exclusive) 13 spoligotypes fallow deer (1

1 spoligotype mouflon (1 exclusive) 3 spoligotypes lynx (none exclusive)

3 spoligotypes wild boar (none

red fox (none exclusive) 2 spoligotypes Iberian lynx (1

22 clusters wild boar (8 exclusive) 13 clusters red deer (3 exclusive) 7 clusters pig (2 exclusive) 3 clusters cattle (1 exclusive)

exclusive)

exclusive)

1 cluster goat

exclusive)

exclusive)

exclusive)

exclusive)

exclusive)

exclusive)

Technique Time frame Genotypes

1998-2001 8

1999-2002 11

1998-2003 14

1992-2004 (4 clusters,

SP 24

SP 1996-2002 21

SP 2002-2007 29

MV 2002-2007 87

SP 1992-2007 252

14 wild boar SP 2005-2006 4

1998-2003 9

Reference Sample

Aranaz *et al*. (1996)

Parra *et al*. (2003)

Aranaz *et al*. (2004)

Gortázar *et al*. (2005)

Parra *et al*. (2005)

de Mendoza *et al*. (2006)

Duarte *et al*. (2008)

Romero *et al*. (2008)

Duarte *et al*. (2009)

Santos *et al*. (2009)

Rodríguez *et al*. (2010)

n (Isolates)

4 wild boar 2 red deer (129 cattle, 44 goat, 1 sheep, 2 cat)

37 wild boar (25 Iberian pig)

33 red deer 62 fallow deer 58 wild boar 3 Iberian lynx (50 cattle)

58 wild boar 19 red deer

112 wild boar 59 red deer (6 cattle, 28 Iberian pig, 2 goat)

11 wild boar 8 red deer (5 cattle)

21 red deer 6 wild boar (258 cattle, 8 goat)

60 wild boar 26 red deer 17 fallow deer 4 Iberian lynx 2 red fox (54 cattle)

13 red deer 4 wild boar (157 cattle, 7 goat)

204 wild boar 141 red deer 229 fallow deer 2 chamois 1 mouflon 6 Iberian lynx 2 red fox 1 badger

SP MV

SP MV

SP MV

SP MV

SP MV


Table 6. Molecular biology studies included in the analysis. SP: spoligotyping, MV: MIRU-VNTR mycobacterial interspersed repetitive units-variable number of tandem repeats.

On the other hand, studies of wild ungulates relying on routine meat inspection for detection of macroscopic tuberculosis-like lesions, do not allow for a reliable estimation of prevalence, which is underestimated in this situation (de Mendonza *et al*., 2006). Nevertheless this type of design allows increasing sample size, which makes them suited for long-term surveillance rather than detailed epidemiological studies (de Mendonza *et al*., 2006) and were mostly used in the first surveys and cross-sectional studies after bTB was detected in wildlife in Iberian Peninsula. The investigations on carnivore species, most of which are not hunted, tend to rely on passive surveillance schemes based on haphazardly found carcasses. This sampling design does not allow to estimate prevalence rates due to extensive sampling bias (e.g. Taylor *et al*., 2002). Targeted sampling in these species has been attempted using serological tests but results should be interpreted with caution since these techniques have not yet been validated in these species.

The number of animals studied is usually adequate to determine prevalence rates with relatively small confidence intervals, at least in the easily available hunted species. The same cannot be said for most studies on protected carnivore species, where the collection of biological samples from a large number of animals is inherently difficult.

Bacteriological culture is the reference test for diagnosing bTB although it is expensive and time-consuming (de Lisle *et al*., 2002). As the financial resources needed to perform bacteriological culture on a large number of samples are scarcely available, most surveys use

Wildlife Tuberculosis: A Systematic Review of the Epidemiology in Iberian Peninsula 287

nowadays occupy almost all Iberian Peninsula (Rosell, 2001). Natural expansion of red deer also occurred but not to such a great extent as in the wild boar case and was much

As suggested by Santos *et al*. (2009) for Portugal, wildlife bTB could be similarly expanding from the historical refuges with a lag comparative to its host's expansion. This lag could be explained by the threshold theory for disease persistence, as reported for other bTB hosts such as the possum *Trichosurus vulpecula* in New Zealand – Lloyd-Smith *et al*., 2005). As wild ungulate populations expanded, densities at the front of the expansion wave were too low (Holland *et al*., 2007) to allow for the persistence of bTB, even if presumably some infected hosts were involved in that expansion event. As a consequence, wildlife bTB initially remained confined to the historical refuges, despite dispersion of infected hosts. As ungulate distribution continued to expand, densities increased in a gradient centred at the historical refuges and eventually reached the threshold level. At that point, bTB, introduced by infected immigrants from the historical refuges, could persist and spread its distribution, a

This hypothesis could be tested by comprehensive geographical spatial analysis of the distribution of bTB in Iberian Peninsula, but the proposed natural expansion pattern has probably been much obscured by translocation and intensive management of ungulates for hunting purposes (Vargas *et al*. 1995; Miguel *et al*. 1999; Castillo *et al*., 2010). In fact, in South-central Spain lack of geographical autocorrelation in prevalence rates was suggested to be due to extensive fencing of intensively-managed big game hunting estates, which impair animal movements (Vicente *et al*., 2006b). On the other hand, wild ungulate translocations for hunting purposes occur frequently and may spread *M. bovis* to areas where it is absent today. Interestingly, *M. bovis* was isolated from wild boar in Portugal in two areas widely out of the known distribution of the disease (Santos *et al*., 2009; Cunha *et al*., 2012), one of which coincides with the release site of red deer originating from a population harbouring the same genotype of *M. bovis*. This provides

circumstantial evidence for the role of translocations on bTB geographical spread.

understanding of bTB occurrence across Iberian Peninsula.

More spatial data of bTb occurrence in Iberian Peninsula is urgently needed. The advent of sensitive, specific, reproducible and cheap serologic tests allows such large-scale research to be conducted, at least for wild boar (Boadella *et al*., 2011). This should improve the

Most risk factors for bTB in wild boar and red deer identified in Iberian Peninsula are host population factors, most of them abundance-related. It is interesting to note that in the wild boar-red deer system, the abundance of each species influences bTB occurrence in the other species, further supporting the multi-host pathogen status of bTB in Iberian Peninsula

The number of risk factors related to management is greater for the red deer (n=5) than for the wild boar (n=1). This suggests that bTB occurrence in red deer populations is more dependent on management practices, while wild boar is competent to act as maintenance host under low-intensity management. This hypothesis could be tested by a case-control study of bTB occurrence in both species across a gradient of intensity of management.

dependent upon translocations (Soriguer, 1998; Acevedo *et al*., 2011).

process seemingly still taking place.

**4.4 Disease determinant factors** 

ecosystems.

other methods (usually gross pathology) as screening tests and only perform bacteriological culture for lesion-positive animals, sometimes as pooled samples. This introduces a bias and it was shown that the sensitivity of gross pathology was 72,2% of that obtained from bacteriology in the wild boar (Santos *et al*., 2010). The same trend has been reported elsewhere for deer (Rohonczy *et al*., 1996; O'Brien *et al*., 2004).
