Potential ecological and epidemiological factors affecting the persistence of classical swine fever in wild boar Sus scrofa populations

Social organization and seasonal population characteristics

Female wild boar typically live in social groups sharing exclusive core areas within their home ranges. These reproductive groups consist of related subadult (> 1 year) and adult (> 2 years) females and their piglets. Groups of up to 30–40 individuals have been reported in non-hunted areas. In hunted populations, group size may be reduced to only 1 or 2 adult females and about 5–10 individuals. This structure exists all year although age-dependent composition of the family group does vary. The groups are spatially restricted to home ranges of about 4–20 km² (Leaper et al., 1999).

Adult males are frequently solitary, but subadult males often roam together in small groups (Fernandez-Llario, Carranza & Hidalgo de Trucios, 1996). During the autumn rutting season, males associate with the female groups for mating. Although it has been demonstrated under experimental conditions that semen of infected male pigs can contain CSFV and the virus can be transmitted to sows and their foetuses (de Smit et al., 1999), evidence is missing for the wild boar.

Subadult females split up and disperse with other subadult females of their natal group to constitute a new social group, although sometimes they may return to their native group in autumn, together with their newborn piglets (Dardallion, 1988). Female groups do not seem to associate with other families, although the respective home ranges may overlap. Thus, group splitting by subadult females extends the spatial range of possible contacts but happens only once a year, normally between September and December. Movement distances of subadult females have been recorded up to 20 km, whereas mean distance was 4.5 km in one study in Sweden. Mean seasonal distances in subadult males were around 17 km, but single individuals were observed to move up to 100 km (Truvé & Lemel, 2003). Hence, the virus might be spread over longer distances in suitable habitat by this process, although no proof is available that infected wild boar are still able to disperse long distances.

Host abundance and density in European populations

Since the early 1950s, wild boar populations have increased both in number and distribution throughout Europe, apparently due to lack of predators, extreme adaptability, artificial feeding and mild winters (Bieber & Ruf, 2005). France, Germany and Italy host several hundreds of thousands of wild boar, and the present population in European Union member states is estimated between 800 000 and 1 million (Saez-Royuela & Telleria, 1986; Artois et al., 2002).

Wild boar densities can vary markedly, both spatially and temporally, depending on factors such as habitat composition, hunting management and density of populations after farrowing in spring. Density is strongly linked to the availability of high-energy foods such as mast in many natural areas where management does not include supplementary feeding (Massei, Genov & Staines, 1996). For example, in the forests of Poland, density is known to vary between 1.5 and 10 individuals per km² (Andrzejewski & Jezierski, 1978).

Transmission and course of infection at the individual level

Within social groups, the virus is transmitted by direct and indirect contact, especially between piglets. Between social groups, transmission can be due to contact during the rutting season, male dispersers and establishment of new social groups (Kaden, 1999). Contact with contaminated excretions and carcasses can also contribute to the spread of CSF in the wild boar population, as the virus survives in the environment under certain conditions for several days or even weeks (Edwards, 2000; Ribbens et al., 2004b; but see Dewulf et al., 2002b).

Insects (flies) have been reported as possible vectors for CSF (Dahle & Liess, 1992), and airborne spread may also occur, as shown in housing experiments (Terpstra, 1987, 1988; Dewulf et al., 2002a). But these factors do not seem to play a significant role.

The incubation period is, on average, 5–10 days (Moennig, 2000). After infection, the clinical courses of CSF range widely from peracute to subclinical to immune outcomes. Several factors have been proposed to influence this variability in pigs, including the age and condition of the infected individual (immune response) and the virus strain's virulence (Tables 1 and 2; Dahle & Liess, 1992; Moennig, 2000; Paton & Greiser-Wilke, 2003). The disease normally kills a higher proportion of young wild boar, because their immune system is not fully developed, whereas in the adult cohort, infections are more likely to result in a lower mortality rate and higher level of immunity.

The acute form is characterized in domestic pigs by high fever, leukopaenia, and haemorrhages of skin and inner organs. Animals showing the acute course die within 30 days. If the animal shows disease symptoms and sheds the virus longer than 30 days, the disease course is classified as chronic. These pigs may repeatedly shed the virus. Lifespans of chronically infected pigs of up to 152 days are reported (Depner, Moennig & Liess, 1997). However, in wild boar only the acute course has so far been described (Mengeling & Cheville, 1968; Dahle & Liess, 1992). Survival of the infected animal is due to the development of long-lasting immunity. This course is classified as transient infection because these animals shed the virus for 1–2 weeks before they become immune. Pigs that have recovered may be protected against CSF for their lifetime (Rossi et al., 2005a).

If the mother sow has antibodies, e.g. after surviving an infection, a certain level of maternally derived antibodies (MDA) is detectable in offspring for the first 8–12 weeks (passive immunization). After these 3 months, low levels of MDA have been detected up to 12 months post partum (K. Depner, personal communication). For such ‘partially protected piglets’ the clinical course of a CSF infection is more often transient than lethal (Depner et al., 2000).

Vertical transmission of CSF leads to prenatal infections, i.e. from the mother to the foetus, which can result in different outcomes depending on the developmental stage of the immune system in a foetus (Artois et al., 2002). Vertical transmission only occurs when at the time of conception the mother sow is antibody negative. Early infections of the pregnant sow normally result in reduced birth probability (abortion, stillbirth), whereas infections close to birth can result in normal and non-infected piglets. But in the middle of the pregnancy period, intrauterine infected piglets are observed (Table 3). Although these animals eventually all die, a persistent infection was observed in experimental studies. The longest infectious period of wild boar piglets was about 40 days (Depner et al., 1995) and up to 11 months for a domestic piglet (Van Oirschot & Terpstra, 1977). Intrauterine infected wild boar piglets might not survive for an equally long time because living conditions in the field are much harder (Moennig et al., 1999).

Prevalence, incidence and transmission at the population level

In continental Europe, CSF is only prevalent in a limited number of foci. At the beginning of the epidemic, all age classes seem to be affected by increased mortality, but the majority of deaths are of young animals. Peaks in mortality have been observed after initial virus introduction, followed by a slow progressive decrease in the infection rate over a long period of time, which is characterized by increasing sero-conversion in adults; that is, surviving adults are getting immune (Artois et al., 2002). Prevalence is highly variable depending on the time after introduction, e.g. 34% virus positive after the beginning of an outbreak and 1.5% one year later, and 5% sero-positive shortly after the outbreak and 55% one year later in an outbreak in Italy (Zanardi et al., 2003).

In previous epizootics it has been observed that whole cohorts died out (Kaden, 1999), but in recent years, the infection was seldom observed to be self-limiting (Ferrari et al., 1998; Fritzemeier et al., 1998). Sometimes the virus was circulating for years (Laddomada et al., 1994; Fritzemeier et al., 1998; Kern et al., 1999), showing an oscillating pattern with periodic peaks, but with a decreasing trend (Laddomada, 2000). Ongoing persistence in the field is usually demonstrated by the detection of antibodies in young boar older than 3 months (i.e. no longer having MDA; Artois et al. 2002). In the following section, we discuss the current hypotheses that might explain why endemic situations have increased.

Intrauterine infected piglets

The mechanism of virus spread through prenatally infected piglets is frequently assumed to play a key role in the persistence of CSF. If intrauterine infections result in a prolonged virus-shedding period compared with naïve post-natal infections in piglets (Van Oirschot & Terpstra, 1977; Liess, 1987; Terpstra, 1988; Kern et al., 1999; Laddomada, 2000), then persistent infection in these piglets might bridge temporal gaps in the infection chain (Kern et al., 1999).

On the other hand, in endemic areas intrauterine infected piglets are unlikely to occur at a substantial rate due to the immunity already acquired by older animals. This suggests that young wild boar persistently infected by transplacental virus transmission do not play a crucial role in the persistence of CSF in wild boar (Moennig et al., 1999; Kaden et al., 2005).

In one study, prenatally infected piglets survived for more than 2 months (see Van Oirschot & Terpstra, 1977). Another study with a virus strain of similarly low virulence indicated that the piglets only survived to a maximum of 8 weeks (Table 3; Meyer et al., 1981). Although we concur with the potential importance of long-lasting infections as drivers for CSF persistence in wild populations, we doubt that intrauterine infected piglets will play that part, unless they survive and continue to be infectious for much longer than identified in previous studies. When asking for persistence, we consider the endemic situation in contrast to an epidemic, where the occurrence of successful prenatal infections is reduced by the immunity of older animals. As the risk of having persistently infected piglets mostly relates to social groups with primary infections, i.e. fully susceptible groups, we suggest that this hypothesis can be rejected as the main driver for CSF persistence under natural conditions.

Transient infections in partially protected piglets

The infection of piglets partially protected up to 1 year by MDA may play a crucial role in the spread of CSF. These pigs may contribute to the maintenance of long-lasting epizootics, because they are not severely affected by the disease and can still move (and carry the virus) long distances (Depner et al., 2000). During group splitting and dispersal, such animals would be a plausible force for spatial viral spread. However, infections of partially protected piglets are shown to be rather transient, i.e. running for at most 3 weeks post infection (Depner et al., 2000). Thus, all infected partially protected piglets will have recovered before they reach the age to disperse. We conclude that the hypothesis per se is not suited to explaining disease persistence in wild boar populations.

Differences in virulence

Sustained transmission is least likely for virulent parasites, i.e. with high transmission rates and short infectious periods. They cannot persist in small host populations unless they also have a reservoir host (Brunner et al., 2004). Although some species like the European rabbit may eventually be infected by CSFV (Chenut et al., 1999), other animals are not reservoirs and even their role as CSFV vectors remains doubtful (Dewulf et al., 2001c; Kaden et al., 2003; Ribbens et al., 2004a). The Virulence Evolution Hypothesis claims that strains of intermediate virulence are the outcome of natural selection to balance between the virus' need to reproduce and the costs of harming the host (Gandon, van Baalen & Jansen, 2002; Mackinnon & Read, 2004).

According to descriptions from the mid-20th century, CSF used to be an acute disease with short incubation period and high mortality rates. The course of the disease has since changed dramatically. Today, the peracute form has disappeared and the acute form no longer results in high mortality rates. In addition, the chronic form of the disease is more common in pigs. The different clinical pictures can be attributed to CSFV strains of varying virulence (Meyers & Thiel, 1996). This striking inter-animal variability in disease outcome of chronic and acute infections in moderately virulent strains might facilitate epidemic as well as endemic situations (Mittelholzer et al., 2000; Uttenthal et al., 2003).

The hypothesis is underpinned by recent genotyping analyses showing a switch from virulent group 1 viruses prior to the 1970s to group 2 viruses (Paton et al., 2000). Moderately virulent viruses of subgroup 2.2 and 2.3 were present in wild boar populations of France, Austria, Italy and Germany, partly in endemic areas (Fritzemeier et al., 1998; Greiser-Wilke et al., 2000; Biagetti et al. 2001; Kaden et al., 2004). In Tuscany, Italy, a decrease in CSFV virulence during an epidemic in wild boar has been described (Ferrari et al., 1998). However, Depner et al. (1997) concluded that the changing clinical picture of CSF observed during the recent epizootic in domestic pigs in Lower Saxony, Germany, was mainly related to host factors such as age and genetic characteristics of the pig, rather than to the virulence of the virus, because even with a highly virulent virus strain, there was a high variability in disease outcome. Whether disease outcome is related to the virulence of the virus or to the condition of the animals is still discussed (Floegel-Niesmann et al., 2003). Genetic resistance, for example, to bovine tuberculosis is suggested in Iberian wild boar (Acevedo-Whitehouse et al., 2005), and although it is not yet recorded, genetic resistance to CSF may also be possible.

The most plausible explanation for virus perpetuation seems to be the change from a highly virulent virus strain to a strain of moderate virulence. What is relevant in terms of population epidemiology is that a higher number of transients and chronics in the population (because of either lower virus lethality or higher host resistance) can be responsible for the persistence of the disease. For simplifying reasons, and because it is still under discussion whether it is due to host or viral traits, we refer to the virulence of the virus in the course of this review as a highly variable disease outcome.

The mechanism behind virus circulation of moderately virulent strains would be that more infections will be transient, guaranteeing sustained reproduction of susceptible individuals due to surviving adults and, in addition, more chronic infections. The latter can shed the virus for up to 6 months in pigs. Thus, chronically infected individuals will occur independent of the state of the epidemic (i.e. the immunity level in the population). However, so far only the acute form has been described in the wild boar (Dahle & Liess, 1992), presumably because identifying chronic disease in the field is not possible at the moment.

Low virulence also increases the probability of having intrauterine infected piglets, as more sows get transient infections. Most studies were conducted with a low virulence virus (Table 3): if low virulence viruses result in more chronic infections with long lifespans in infectious individuals, the detected long life times of prenatally infected piglets can also simply be due to mothers infected with a less-virulent strain.

Population characteristics and dispersal

The social structure of wild boar populations may favour the spread of diseases. Only brief contact is required, and that level of contact is very high within social groups or when aggregating around artificial feeding or watering sites (e.g. Vicente et al., 2004, 2005). After a population crash, a typical aspect of wild boar populations under natural conditions (Okarma et al., 1995; Jedrzejewska et al., 1997; Massei et al., 1997), is a high population production, which contributes to the rapid establishment of a newly susceptible generation. Also, wild boar show intense responses to food pulses (mast years, artificial feeding) with increased reproduction (Bieber & Ruf, 2005). As newborns are numerous and also the most vulnerable age class for CSF, these cycles might mediate the continued spread in local areas (Laddomada, 2000). Further, seasonal dispersal may enhance spread between regions. During the dispersal season, previously infected social groups with a number of susceptible offspring can come into contact with infected animals. Thus, dispersal may link the ‘susceptible patches’ even within the same population (Laddomada, 2000). Therefore, virus persistence in wild boar populations could be a result of a fast re-establishment of a new susceptible population as well as immigration.

Population size (spatial extent) and density

The increase in wild boar populations has been linked to virus perpetuation. It seems striking that the cases of endemic situations in recent years coincide with the population explosion of wild boar at the beginning of the 1950 (Jedrzejewska et al., 1997). Increase in wild boar numbers is not only related to an increase in wild boar densities, but also to an extended spatial range. Here, we refer to population size as the spatial extent a connected population covers and to density as the number of individuals living in a certain area, and review the effects of size and density separately.

(a) Population size. In general, theory predicts that a large population contributes to long-lasting epizootics, as it takes a longer time until the whole population has been completely exposed to infection (Mollison & Levin, 1995; Hudson et al., 2002). If the landscape is large or the disease spreads very slowly, a fully susceptible new generation can be raised ‘behind the primary epidemic wave’. Thus, a second infection wave can occur, although with lower peaks of infection, because many adult individuals are already immune (Swinton et al., 2002). This corresponds with the oscillating pattern found during an epidemic in France, with peaks of disease occurring every 2 years, but with a trend towards a decreased size of the peaks (Laddomada, 2000).

However, we think that this hypothesis does not work well with a highly virulent CSF virus for locally restricted areas. Therein, the depletion of whole groups (Kaden, 1999) makes reproduction and thus secondary infections in the population unlikely, and the infection might become self-limiting after the primary outbreak. However, the virus has been demonstrated to circulate in small foci (Laddomada et al., 1994), indicating that the population size hypothesis is not sufficient to explain CSF persistence.

(b) Population density. Variability in population density is one of the prime factors for explaining differences in disease persistence in epidemics and particularly for CSF virus cycling, assuming that a high wild boar density can contribute to keep the CSF virus transmission chain connected (Kern et al., 1999; Rossi et al., 2005a,b). For many infectious diseases, transmission is assumed to be a density-dependent process. Below a critical host population density, the contact rate between infected and susceptible hosts is too low to ensure transmission, and so the pathogen goes extinct (Brunner et al., 2004). Thus, epidemiological studies often refer to a threshold density below which the disease is likely to fade out. For CSF, the density threshold was calculated with deterministic and stochastic models, and for different populations, to be about 1 individual per km² (range 0.6–1.1) (Guberti et al., 1998) or 6 individuals per km² (range 5–10) (Hone, Pech & Yip, 1992). Nevertheless, CSF has persisted for many years in low-density populations as well as in high-density populations (Artois et al., 2002). This has also been found for rabies in foxes, and recent modelling efforts for territorial wildlife species have found evidence that density alone does not determine disease fade-out (Sterner & Smith, 2006). This may be due to the fact that the density concept is related to a crowding of individuals per area, which does not directly alter the temporal dynamics of the epidemic, i.e. the bridging of the ‘troughs’ (Lloyd-Smith et al., 2005) between epidemic peaks via the replenishment of the susceptible pool through reproduction.

However, local differences in spatial and temporal aggregation may explain CSF virus transmission and perpetuation, as has been linked with a number of other infectious diseases in wild boar, e.g. Circovirus (Vicente et al., 2004) or Aujetzky's disease (Vicente et al., 2005). These spatio-temporal aggregations may be due to habitat fragmentation but most obviously to artificial feeding and, in southern Europe, watering (Gortázar et al., 2006). Artificial feeding or watering also has a positive feedback on population growth and thus again on density. There is another side effect of artificial feeding: when wild boar are in good condition due to high food availability, it is more probable that chronic or transient infections will occur, because the disease outcome is also related to the host condition.

In summary, large host population numbers, either occupying large connected areas or living at high local densities, increase the likelihood of disease persistence simply because more rare events happen (like chronic disease courses, long-distance dispersal) and more immune animals can reproduce. Thus, with persistence being determined by relationships between host numbers and the timescales of demography and transmission, there is a gradual dependence on host numbers (Lloyd-Smith et al., 2005). Therefore, a large host number might support the persistence mechanism of CSF, but is rather not the key concept allowing discrimination between persistence and fade-out.