Combining hunting and intensive carcass removal to eradicate African swine fever from wild boar populations.

African Swine Fever (ASF) is a highly lethal viral disease, which affects different species of wild and domestic suids. After its human-caused introduction in Georgia in 2007, the ASF virus has found a new ecological reservoir in the large and continuous wild boar (Sus scrofa) populations of Eurasia, spreading both eastward and westward. ASF has also breached into the intensive pork meat production system. Although the disease has no zoonotic potential, its consequences on wild boar populations and the economic losses for the pig industry have been dramatic. As no vaccine or effective medical treatment is available to reliably protect wild boar or domestic pigs against ASF, eradication efforts are mainly based on intensive wild boar hunting and on removing a significant portion of the infected wild boar carcasses, which are the main environmental virus reservoir. Both strategies have produced poor results, so far, and ASF is becoming endemic. We compared wild boar hunting and carcass removal as alternative and combined strategies for the eradication of ASF in its endemic state, using a spatially explicit individual-based model, which incorporated the demography and spatial dynamics of a wild boar population, the spatial epidemiology of ASF in its endemic phase, and a management system acting for the eradication of the disease from the population. When no eradication effort was simulated, ASF exhibited a clear and strong tendency to persist and remain endemic in the wild boar population. Both hunting and carcass removal, when used alone, provided either a low power to remove the virus from the population, or required unrealistic field effort. The best performing scenario corresponded to the combined use of a 30% annual hunting rate and of an intensive carcass removal, during a 2-month period in late winter (February-March). Eradicating ASF from wild boar populations remains a hard task. Managers should promote a drastic increase in the effort dedicated to systematically identify and remove as many infected wild boar carcasses as possible from the affected areas, with at least 5-15 carcasses removed for each 100 hunted wild boar.

Introduction

In recent years, a growing concern is emerging about the potential by diseases in wild animal populations to affect human health, animal productivity in agricultural systems, and global biodiversity (Jones et al., 2008). Many infectious diseases in humans, such as HIV/AIDS, SARS, and more recently COVID-19, are zoonoses derived from wild animal populations (Adnan et al., 2020). Moreover, wildlife populations can represent important reservoirs for economically significant livestock diseases (Craft, 2015), whose appearance into the food production system can cause large economic losses (Pike et al., 2014). As the risk of new spillover events from wild populations to domestic species is increasing globally, due to extensive land use change, deforestation, and environmental encroachment (Hassell et al., 2017), the potential impact of wildlife pathogens on human livelihoods becomes more and more a relevant factor when dealing with natural resource management (Di Marco et al., 2020).

African swine fever (ASF) is a glaring case of such a risk. The disease is caused by a highly virulent virus of the Asfaviridae family, which affects different species of wild suids. Historically, ASF was distributed in eastern and southern Africa with 24 different genotypes, where it was (and still is) maintained in an ancient sylvatic cycle among warthogs (Phacochoerus africanus) and the soft ticks of the genus Ornithodoros (Jori et al., 2013), although the role of human-mediated transmission and of other socio-economic factors is also relevant (Penrith et al., 2019). In 2007, though, a human-caused new introduction of ASF virus genotype II was observed in the ex-soviet republic of Georgia, which affected both domestic pigs and wild boar (Sus scrofa), causing an important disease with almost 100% lethality (Blome et al., 2013). From there, the virus slowly spread both eastward and westward (Penrith, 2020), finding a new ecological reservoir in the large and continuous wild boar populations of Eurasia. Nowadays, the ASF virus genotype II has its most western European front in Germany, while the virus has progressively infected almost all countries in East and South East Asia including China (2018), Philippines, Papua New Guinea and Timor-Leste (Sauter-louis et al., 2021a). In 2021, the ASF genotype II has reached Santo Domingo, spreading to both the Dominican Republic and Haiti.

In Eurasia, shortly after its appearance, ASF has breached into the intensive food production system of pig farming, becoming both an ecological and an economic concern (Penrith, 2020). Although the disease has no zoonotic potential (Alonso et al., 2018), it determines huge direct and indirect economic losses (Pitts and Whitnall, 2019). The pig industry sector suffered an acute fall in meat production and a sharp increase in prices; in 2018, China, which accounts for 45% of the world’s pig meat production, has suppressed 25–50% of its stock due to ASF infections, while pork meat prices have increased globally by 60–85% (Tian and Cramon-taubadel, 2020).

In most wild boar populations in Eurasia, the onset of ASF was characterized by an epidemic wave, during which population abundance was reduced up to 85% (Morelle et al., 2020). The initial wave was invariably followed by a long-term endemic status, such as in Poland and the Baltics, which have been infected since 2014. As no vaccine or effective medical treatment is available to reliably protect wild boar or domestic pigs against ASF (Sauter-louis et al., 2021a), two main strategies have been carried out to eradicate the virus from the European infected wild boar populations: a) fencing of the affected area to prevent any further geographical spread, followed by the active search and removal of wild boar infected carcasses (Boklund et al., 2018); b) the progressive reduction of population density and virus incidence through targeted hunting. Both the strategies were strengthened by the implementation of strict biosecurity measures during wild boar hunting and in pig farms, to minimize the probability of human mediated spread of the virus and to prevent its entrance in pig farms, in the pork production system, and in ASF free forests (Chenais et al., 2019).

Despite such efforts, ASF has continued to spread in most of the European wild boar infected countries. Field evidence and ASF modelling studies have shown that the disease can persist with a very low prevalence through three main mechanisms: live infected wild boar, convalescent but still infectious individuals, and the high tenacity of the virus in carcasses and in the environment (Pepin et al., 2020). Virus persistence occurs even when wild boar density is kept low through intensive hunting, with the consequence that ASF can remain substantially invisible to surveillance, until the conditions for a new epidemic outbreak arise.

Eradicating endemic ASF has proven to be a challenging, often unsuccessful task. Intensive wild boar hunting is aimed at keeping population density below the theoretical contact threshold allowing virus effective maintenance (More et al., 2018). Such strategy is limited by the fact that disease population thresholds in wild mammals are usually not abrupt, difficult to estimate, and prohibitively hard to reach in practice (Lloyd-Smith et al., 2005). The ASF eradication strategy based on carcass removal has also been relatively unsuccessful (Nielsen et al., 2021). Despite its strong theoretical basis, because it reduces contact rates between susceptible individuals and the main environmental virus reservoir, this strategy has resulted in a relatively low number of removed carcasses, so far (Nielsen et al., 2021), with the detected carcasses likely being a small proportion (< 5%) of all the carcasses expected in the landscape during the endemic period (Lange et al., 2021). As the logistic and human resources available for eradication programs are often limited, understanding which strategy or combination of strategies is more likely to be effective in ASF eradication is a fundamental prerequisite to properly allocate resources, design policies, and evaluate them in an adaptive management framework. Several expert opinions (Lange et al., 2018, Nielsen et al., 2021) and research studies (Croft et al., 2020, O’Neill et al., 2020) have qualitatively evaluated the benefits and challenges of different approaches, but a formal, quantitative evaluation of the eradication chances associated to different management actions has not been fully performed, so far.

In this paper we try and fill this gap by formally comparing wild boar hunting and carcass removal as alternative or combined strategies for the eradication of ASF in its endemic state. We developed a spatially explicit individual-based model (IBM), which incorporated three main components: i) the demography and spatial dynamics of a wild boar population; ii) the spatial epidemiology of ASF in its endemic phase; iii) a management system acting for the eradication of the disease from the population through wild boar hunting and carcass search and removal. We discuss the implications of our study for a better understating of the ASF spatio-temporal dynamics in its endemic state, and provide applied management indications on how to increase the chances of ASF eradication in different ecological contexts.

Methods

Demographic model

We built and ran the IBM using the software Netlogo 6.1.1 (Wilensky, 1999). To define the demographic component of the model, we first constructed a simulated squared landscape of 8100 km2, in which all model processes occurred. Then, we divided the landscape into 900 3 × 3 km cells, so that each cell corresponded to the realistic size of a wild boar’s core home range (Leaper et al., 1999). We initially populated the simulated landscape with 8100 wild boar individuals, corresponding to a uniform density of one wild boar / km2. The initial population had a 1:1 sex ratio and comprised 60% juveniles, 20% yearlings and 20% older individuals, corresponding to the post-reproductive stable-stage distribution (Bieber and Ruf, 2005).

In the demographic part of the model, we only considered natural mortality as a possible risk for wild boar. We applied a 0.18 annual rate to juveniles and 0.12 to the other two age classes (Toïgo et al., 2008). We also modelled year-to-year variation in the population reproductive output though the following age and density-dependent function (Eq. 1), controlling the proportion of females producing a litter each year:in which R was the probability to reproduce for a female of age class i, α was proportion of females of age class i giving birth in the absence of any density-dependence, β was the intensity of the density dependence negative effect, N was the local wild boar abundance, measured inside each landscape cell, and K was the local carrying capacity. We defined parameters α and β to maximise the fit between the density-dependent function and the fecundity estimates reported by Bieber and Ruf (2005) in good, intermediate, and poor environmental conditions (see Table 1). We set carrying capacity K at 4 wild boar / km2. Based on the above-defined reproduction probability, at the beginning of the reproductive season (April 1st) the model assigned each female wild boar with a reproductive state. Then, a delivery day was generated, based on a uniform distribution ranging 1–60, thus allowing births to be uniformly distributed during the 2-month reproductive season. We set litter size to 4 piglets for juveniles, 5 for yearlings, 6 for adults (Bieber and Ruf, 2005) with a 1:1 sex ratio.

Table 1 –

Summary of the main parameters used to run the individual-based model of ASF endemic persistence in wild boar.

Parameter	Description	Value	Source
N0	Initial population size	8100	–
D0	Initial density (wild boar / km2)	1	–
A0	Initial age-distribution (juveniles, yearlings, adults)	0.6, 0.2, 0.2	Bieber and Ruf (2005)
Pd	Direct transmission probability (infected – susceptible)	0.0035	Gervasi et al. (2021)
Pc	Carcass transmission probability	0.00016	Gervasi et al. (2021)
Ps	Convalescent transmission probability	0.00038	Gervasi et al. (2021)
ε	Incubation time (days)	3	Blome et al. (2013)
γ	Disease lethality	0.946	Gervasi et al. (2021)
I	Carcass infectious period (days)	85	Gervasi et al. (2021)
χ	Convalescents infectious period (days)	77	Gervasi et al. (2021)
M	Natural mortality rate (juveniles, yearlings, adults)	0.18, 0.12, 0.12	Toigo et al. 2008
α	Reproduction probability (juveniles, yearlings, adults)	0.5, 0.8, 0.9	Bieber and Ruf (2005)
β	Intensity of the density-dependence in reproduction	5.8	–
K	Carrying capacity (wild boar / km2)	4	–
L	Litter size (juveniles, yearlings, adults)	4, 5, 6	Bieber and Ruf (2005)
d	Dispersal probability (females, males)	0.4, 0.7	Truvé et al. (2004)
H0	Annual hunting rate	0 – 0.5	–
C0	Probability to find and remove a carcass before decomposition	0 – 0.95	–

The piglets coordinates initially corresponded to those of their mother, but were modified in the following year by the dispersal process. At the beginning of the dispersal season, which occurred between June 1st and August 10th, all yearling individuals were assigned a dispersal state, generated through a sex-specific dispersal probability (0.7 for males, 0.4 for females; Truvé et al., 2004). Dispersal duration ranged from 1 to 14 days for males and from 1 to 7 days for females (Truvé et al., 2004). Each day, dispersing individuals moved away from their current cell to the neighbouring cell with the lowest wild boar density, thus mimicking the effort to avoid intra-specific competition for resources. They finally settled in a new home range at the end of their dispersing period.

Epidemiological model

To develop the epidemiological component of the IBM, we added a compartmental structure to the wild boar simulated population, which allowed to mimic the characteristics of a classical SEIR epidemiological model (Anderson and May, 1992). To this aim, we initially included 92% of the wild boar population in the susceptible (S) compartment, which comprised the healthy individuals available for infection; 1% of the population was included in the infected (I) compartment, thus defining the initial endemic ASF prevalence; finally, we included 7% of the population in the recovered (R) compartment, which comprised those seropositive individuals which survived the disease and became immune. We derived the endemic prevalence and seroprevalence values from the ASF surveillance data reported in Estonia and Latvia during the endemic phase of the disease, in years 2018–2020 (Nielsen et al., 2021).

We modelled virus transmission through three different mechanisms. A susceptible individual could experience infection after contact with: i) an infected individual in the acute phase of the disease; ii) the infected carcass of a wild boar dead due to ASF; iii) a recovered wild boar still shedding the ASF virus during its convalescence period (Fig. 1). Moreover, infection could occur either between individuals occupying the same landscape cell (and likely belonging to the same social group) or between wild boar living in neighbouring cells, with the infection probability being ten times higher in the first than in the second case (Gervasi and Guberti, 2021). After infection, individuals were first moved into the exposed (E) compartment, which corresponded to the 3-day period of incubation and latency (Blome et al., 2013), then transferred into the infected one. There, a wild boar could transmit the disease with probability P and had a γ = 0.945 probability to die because of the ASF acute infection within a 5-day period (Gervasi and Guberti, 2021). The individuals which did not survive the disease were transferred into the infectious carcass (C) compartment, where they remained for a period of 85 days in winter and 42 days in summer (Gervasi and Guberti, 2021), during which they could act as a source of additional infections with probability P. After such period, they were removed from the model, thus mimicking full carcass decomposition. The portion of individuals which survived ASF were moved to the recovered compartment, in which they were immediately immune to a potential reinfection, but could still transmit the disease to susceptible wild boar for a period of 77 days with probability P (Gervasi and Guberti, 2021), after which they could neither be re-infected, nor transmit the disease to others. The list of all parameters involved in virus transmission is shown in Table 1, whereas a detailed description of how model parameters were estimated can be derived from Gervasi and Guberti (2021). The sensitivity of ASF persistence to changes in model parameters can also be found in Gervasi and Guberti (2021).

Fig. 1

Schematic representation of the individual-based model of wild boar demography and African swine fever epidemiology, used to test the effectiveness of wild boar hunting and carcass removal as possible disease eradication strategies. Dashed lines indicate infection routes, whereas solid lines indicate demographic processes (reproduction, dispersal and natural death).

Management model

To simulate the effort of ASF eradication, we completed the model with two routines which allowed to remove wild boar alive individuals and wild boar infected carcasses from the simulated landscape, thus mimicking hunting and carcass removal as alternative management options. Wild boar hunting lasted 150 days, during the period 15th September – 15th February, and was controlled by an individual daily probability H to be shot, which we kept equal among both sexes and all age classes. To calculate H, we set an annual hunting rate H and then calculated the daily probability to be hunted as:in which n was the number of days in a hunting season. Hunted individuals were removed from the population and added to an additional H compartment, which had the only purpose to keep track of the number of wild boar shot each day. Similarly, the search and removal of infected carcasses was included in the model through a probability for each carcass to be found and removed before its natural decomposition, expressed by the parameter C. The daily probability to find and remove a carcass (C) was then calculated adapting Eq. (2), in which n was equal to 85 in winter and 42 in summer. Once an infected carcass was found, it was removed from the simulated landscape and transferred to compartment RC, which allowed to monitor the number of infected carcasses found and removed each day.

Management scenarios

To compare the effectiveness of different ASF eradication strategies, we constructed four main management scenarios, corresponding to different approaches and priorities. In a first scenario, we simulated the “sit and wait” approach to ASF control, in which we examined the temporal and spatial dynamics of the disease in its endemic phase, in the absence of any management action. To this aim, we ran the demographic and epidemiological parts of the model, but removed both hunting and carcass removal. This scenario also served as a baseline for the subsequent evaluation of the other scenarios, in which different management actions were simulated. In the second scenario, we estimated the likelihood of ASF eradication in case all the effort was dedicated to wild boar hunting and no action was taken to search and remove infected carcasses. We set parameter C to zero and simulated parameter H in the range 0.1–0.5, thus mimicking an increasing effort by managers to lower wild boar density and remove potentially infectious individuals. In the third scenario we simulated the opposite situation, in which all the effort was dedicated to remove infected carcasses all year round, while hunting was stopped. To this aim we set the parameter H = 0 and simulated and increasing probability to find and remove infected carcasses, with parameter C ranging from 0.25 to 0.75. Finally, we ran a fourth scenario, in which both hunting and carcass removal were used as ASF eradication tools. This scenario involved a larger number of parameter combinations, to mimic a gradient of management priorities, in which most of the effort was dedicated either to hunting or to carcass removal. We simulated this gradient allowing both H and C to vary between 0.1 and 0.5 (Table 2).

Table 2

Summary output of 12 simulated scenarios, mimicking different African swine fever (ASF) eradication strategies in wild boar. The scenarios differ in terms of the effort devoted to wild boar culling and carcass removal, and were derived from an individual-based model of wild boar demography and ASF epidemiology.

Scenario	Hunting rate	Carcass detection rate	Population density (wild boar / km2)			Yearly removed carcasses / 100 km2			Yearly hunted wild boar /100 km2			ASF eradication probability
			Mean	LCI	UCI	Mean	LCI	UCI	Mean	LCI	UCI	1-year	3-year	5-year
Sit and wait	0	0	1.1	1.0	1.3	0	–	–	0	–	–	0.00	0.01	0.01
Hunting only	0.1	0	1.2	0.8	1.4	0	–	–	15.6	11.9	24.7	0.00	0.00	0.01
	0.3	0	0.9	0.7	1.1	0	–	–	31.9	27.6	47.3	0.05	0.10	0.11
	0.5	0	0.6	0.4	0.7	0	–	–	57.4	41.9	60.4	0.34	0.66	0.68
Carcass removal only	0	0.25	1.3	1.1	1.5	15.5	11.9	18.3	0	–	–	0.00	0.01	0.03
	0	0.5	1.4	1.2	1.6	29.1	23.1	34.8	0	–	–	0.48	0.58	0.61
	0	0.75	1.5	1.3	1.6	35.8	25.3	39.8	0	–	–	0.68	0.94	0.94
Mixed strategy	0.1	0.5	1.2	1.0	1.4	25.6	19.5	29.2	15.5	14.4	17.3	0.01	0.33	0.35
	0.2	0.4	1.1	0.9	1.3	14.6	13.9	28.4	30.1	28.0	33.9	0.13	0.69	0.71
	0.3	0.3	1.0	0.9	1.1	9.0	8.9	9.1	46.7	45.2	58.1	0.57	0.97	0.99
	0.4	0.2	0.9	0.8	1.0	8.3	7.1	9.7	53.9	51.4	62.2	0.03	0.95	0.96
	0.5	0.1	0.7	0.5	0.9	5.1	3.4	6.9	56.4	49.1	59.8	0.08	0.71	0.73

Analysis of model results

We ran each scenario and combination of parameters over five years and 100 iterations, with the number of model runs for each scenario being a compromise between the expected accuracy of the resulting estimates and processing time. For each simulated day, we recorded the ASF prevalence and seroprevalence, the number of infected carcasses in the landscape, the number of hunted wild boar and removed carcasses, and population size. We stopped model running once the ASF virus disappeared from the population, i.e. when there were no more infected or convalescent wild boar, and no more infected carcasses. Such situation corresponded to a successful ASF eradication event.

After completing all simulations, for each scenario we summarized the demographic and epidemiologic trends over the five-year simulated period. We also reported the proportion of model runs in which the disease was successfully eradicated at one, three and five years from the beginning of management efforts. This represented the most direct estimate of how effective a given management strategy was in ASF eradication.

Year round vs. intensive carcass removal

In all previous scenarios, we tested the effectiveness of a carcass removal strategy in which searching effort was uniformly distributed during the whole year. As both carcass density and ASF prevalence are known to exhibit seasonal fluctuations (Depner et al., 2017, Frant et al., 2021), we also tested if an alternative strategy, based on a shorter but more intensive period of carcass removal, could provide higher chances of ASF eradication or reduce the logistic burden of a continuous effort. To this aim, we pooled all simulated scenarios and selected the model runs which resulted in ASF eradication. Then, from these runs we plotted the day on which the ASF virus disappeared from the population, to verify in which period of the year eradication was more likely to be achieved. This allowed us to run an additional management scenario, in which carcass search and removal was based on a more intensive effort, but limited to the most favourable period of the year for ASF eradication (see Results). As we set C = 0.95, we essentially tested the performance of a management strategy in which almost all infected carcasses were found and removed from the landscape in a short period of the year. We evaluated the effectiveness of such a strategy in a range of increasing hunting effort (H = 0.1–0.5).

Finally, after identifying the most effective scenario in terms of eradication probabilities, we also ran it in a range of different wild boar density (0.5 – 2.0 wild boar / km2) and ASF prevalence (0.5 – 2.0%), to assess its performance when applied to populations living in different ecological and epidemiological contexts. As hunting and carcass removal probabilities can be difficult parameters to be estimated in the field, for each density scenario we also calculated the ratio between the number of removed carcasses and the number of hunted wild boar, as a more practical index of the relative effort needed for each of the two eradication strategies.

Results

Sit and wait” strategy

When no eradication effort was simulated (no hunting and no carcass removal), ASF exhibited a clear and strong tendency to persist and remain endemic in the wild boar population for the whole simulated period. The disease was still present in the population in 100% of the iterations after one year and in 99% of the cases after three and five years (Table 2). The ASF virus endemic prevalence remained below or around 0.01 during the 5-year period, exhibiting regular seasonal fluctuations: the highest prevalence was observed in summer, after the recruitment of new piglets in the populations, whereas the lowest prevalence coincided with late winter (Fig. 2a). The density of infected carcasses in the landscape exhibited similar annual fluctuations (Fig. 2b) around an average of 10.1 carcasses / 100 km2 (95% CIs = 4.4 – 18.2). The endemic persistence of the ASF virus also prevented the demographic recovery of the wild boar population (Fig. 2c), as wild boar density remained constant around 1.1 individuals / km2 (95% CIs = 1.0 – 1.3).

Fig. 2

Daily trends in African swine fever (ASF) endemic prevalence (a), infected carcass density (b), and population density under a scenario in which no management action is taken to eradicate the disease. The results were derived from an individual-based model of wild boar demography and ASF epidemiology.

Hunting as the only eradication tool

Hunting as the only strategy to eradicate ASF exhibited a low power to remove the virus from the population when the simulated hunting rates were kept at low or medium levels. ASF persistence approached 100% when H was set to 0.1 (Table 2), showing that a moderate hunting pressure corresponded to no hunting at all, in terms of eradication chances. When hunting rate was raised at 30% of the population (H = 0.3), the eradication probability was 5% in the first year and 11% within five years (Table 2). A further increase in hunting pressure, though, substantially improved the chances to eradicate ASF: by removing each year 50% of the population, the model provided a 34% chance to eradicate the virus within one year and 66% in three years (Table 2). Such a high hunting effort corresponded to shooting in average 57.4 wild boar / 100 km2 each year, and to lowering wild boar population density to 0.6 / km2.

Carcass removal as the only eradication tool

When 25% or less of the infected carcasses were removed from the population, the management system exhibited no power in eradicating ASF, with persistence probabilities approach 100% at all time intervals (Table 2). Reasonable eradication probabilities were obtained only for carcass removal rates equal or higher than 50%. In particular, the scenario in which C was set to 0.75 provided a 0.68 probability to eradicate ASF within the first year and 0.94 within three years (Table 2). The eradication goal was obtained by removing each year an average of 35.8 infected carcasses / 100 km2 (95% CIs = 25.3 −39.8). Differently from what observed in hunting-based efforts, ASF eradication through carcass removal occurred while allowing a wild boar population recovery. The average density during the 5-year period, in fact, increased from 1.0 to 1.5 individuals / km2 (95% CIs = 1.3 – 1.6; Table 2).

Mixed strategy

Among the simulations in which both hunting and carcass removal were used simultaneously as eradication strategies, the scenarios with a 30% annual hunting rate and a 30% probability to detect and remove infected carcasses exhibited the highest chances of eradicating ASF (Fig. 3). Such scenario was especially advantageous in terms of the time necessary to remove the virus from the population, as eradication occurred already in the first year in 57% of the cases, and within three years in 97% of the iterations (Table 2). All the other combinations of parameters provided less than 15% chances to eradicate ASF already during the first year (Fig. 3). In the best scenario, ASF prevalence and carcass density still exhibited a small increase in prevalence during the first summer after model start, but then faded in subsequent years without showing any ability to regain the epidemiological patterns observed before management interventions (Fig. 4a, b). ASF eradication through a mixed strategy also occurred while allowing a progressive recovery in wild boar density, which approached 1.6 individuals / km2 at the end of the 5-year simulated period (Fig. 4c). The number of yearly removed carcasses and of hunted wild boar in each scenario are reported in Table. S1 in the Online Supporting Information.

Fig. 3

African swine fever (ASF) eradication probabilities under a set of management scenarios in which both wild boar hunting and carcass removal are used as eradication tools. Eradication probabilities are shown at one (continuous line) and three years (dashed line) after the start of the simulated eradication campaign.

Fig. 4

Daily trends in African swine fever (ASF) endemic prevalence (a), infected carcass density (b), and population density under a scenario in which a 0.3 annual hunting rate and 0.3 year-round carcass detection rate were simulated. The results were derived from an individual-based model of wild boar demography and ASF epidemiology.

Intensive carcass removal

The plot of ASF extinction dates in all simulated scenarios showed that the time of the year with the highest chances to eradicate the virus corresponded to a 2-month period (February-March) in late winter, between the end of the hunting season and the beginning of the reproductive period (Fig. 5). Therefore, we tested the scenario of a short but intensive carcass removal during such 2-month period.

Fig. 5

Frequency distribution of the days of year (1 = Jan 1st; 365 = Dec 31st) during which African swine fever (ASF) went extinct in all simulated scenarios. Coloured bounding boxes indicate wild boar hunting and reproduction seasons. The results were derived from an individual-based model of wild boar demography and ASF epidemiology.

Simulation results showed that such a scenario exhibited about the same performance as the scenarios based on a year-round effort, but with a reduced total effort. When coupled with a 0.3 annual hunting rate, the intensive carcass removal provided 46% chances to eradicate ASF during the first year and 96% in three years (Table 3), a similar performance to the one exhibited by the corresponding year-round effort scenario (Table 3). Conversely, the year-round effort required the annual removal of 9.0 carcasses / 100 km2, whereas the intensive effort scenario corresponded to only 4.7 (Table 3), with a 48% reduction in field effort.

Table 3

Comparison of African swine fever (ASF) eradication probabilities under two different carcass removal strategies, one consisting in a year-round effort with low detection probability (p = 0.3), one in a 2-month intensive effort with high detection probability (p = 0.95). The results, expressed in terms of one-, three- and five-year eradication probabilities, are provided for a range of annual hunting rates.

Hunting rate
	Yearly removed carcasses / 100 km2)			ASF eradication probability			Yearly removed carcasses / 100 km2)			ASF eradication probability
	Mean	LCI	UCI	1-year	3-year	5-year	Mean	LCI	UCI	1-year	3-year	5-year
0	13.6	10.9	15.1	0.00	0.02	0.02	7.3	5.7	8.1	0.04	0.03	0.03
0.1	12.9	10.4	14.3	0.01	0.04	0.05	6.8	5.4	7.9	0.12	0.41	0.45
0.2	11.6	10.9	13.6	0.03	0.59	0.61	6.1	4.7	7.6	0.30	0.32	0.34
0.3	9.0	8.9	9.1	0.57	0.97	0.99	4.7	3.0	6.5	0.46	0.95	0.96
0.4	4.8	4.0	5.7	0.68	0.98	0.99	2.5	1.9	2.7	0.53	0.96	0.99
0.5	3.1	2.3	4.1	0.74	0.99	0.99	1.6	1.4	1.9	0.67	0.98	0.99

When running the 2-month carcass removal scenario in a range of varying wild boar density and ASF prevalence values, the model showed that the relative importance of carcass removal increased when reducing wild boar density and when increasing ASF prevalence (Fig. 6). At 0.5 wild boar / km2 and 2% virus prevalence, 14.0 removed carcasses for each 100 shot wild boar (95% CIs = 11.0 – 16.1) were necessary to achieve eradication; at 2.0 wild boar / km2 and 0.5% ASF prevalence, such ratio was reduced to 5.4 (95% CIs = 3.7 – 6.7). All values are provided in Table S2.

Fig. 6

Relationship between the number of detected carcasses and the number of hunted wild boar in a range of increasing wild boar population densities and ASF prevalence values, under the best performing African swine fever (ASF) eradication scenario (hunting rate = 0.3, carcass detection rate = 0.3). The results were derived from an individual-based model of wild boar demography and ASF epidemiology.

Discussion

Highly lethal acute infections are expected to generate self-limiting epidemics, whose massive associated mortality rapidly restricts the pool of susceptible hosts, leading to a progressive fade-out (Lloyd-Smith et al., 2005). Because of its long-term reservoir of widespread wild boar infected carcasses, ASF has clearly contradicted such expectation, maintaining an endemic condition in most of the affected countries, several years after its first appearance, despite its high associated mortality rates (Nurmoja et al., 2017). In this sense, the results of our IBM fitted closely with the evidence resulting from the European ASF surveillance data (Boklund et al., 2018, Nielsen et al., 2021), while providing a possible mechanistic explanation for the observed patterns.

The results of the “sit and wait” scenario (Table 2) showed that ASF is bound to persist in its endemic state, if no management action is undertaken to remove it from wild boar populations. Moreover, persistence probabilities remained high, almost certain, even five years after the beginning of the endemic phase (Table 2), suggesting that stochastic fluctuations are not likely to increase fade-out chances over time. This represents a serious issue from a management point of view. Since ASF endemic persistence occurs at very low prevalence (< 1%), as confirmed by our model results and by several national surveillance programs (Boklund et al., 2018, Nielsen et al., 2021), the disease is likely to remain substantially invisible to monitoring for several years after the end of the acute epidemic phase. During such “latent” period, which can generate a false sense of reduced impact on the wild boar population, sporadic human-related transmission events or reduced biosecurity measures in pig farms can cause new outbreaks in previously unaffected areas, with serious ecological and economic consequences. Therefore, not actively managing ASF in its endemic period should not be considered as an effective strategy, even when surveillance programs fail to report new cases for a relatively long period of time.

In most of the simulated scenarios, both hunting and carcass removal strategies exhibited a low efficacy in reducing ASF persistence probabilities, when applied as stand-alone eradication tools. In the few cases in which they provided non-negligible eradication chances, their good performance was associated with very demanding, almost unrealistic requirements in terms of effort, which make their practical applicability rather challenging. In the case of carcass removal, a minimum of 30 yearly removed carcasses / 100 km2 was necessary to generate eradication probabilities > 0.5 within three years (Table 2). This represents an unlikely increase in carcass detection rates, if we consider that less than 0.5 carcasses / 100 km2 have been detected and removed in Latvia during the period 2017–2019 (Oļševskis et al., 2020). In the case of hunting as a stand-alone strategy, eradication required to remove at least 50% of the wild boar population each year for several years in a row, and to halve the already low endemic wild boar density (Table 2). As the quota-filling performance by hunters and its predictability over time are expected to decrease for decreasing game densities (Bischof et al., 2012), such high hunting rate objectives should be considered as prohibitive under most field and logistic conditions. Moreover, recent studies highlight the potential trade-off between combating disease transmission within a population using hunting and the risk of causing a further geographic spread (Mysterud et al., 2020). High hunting pressure can cause a progressive disruption of social structures and increase individual movements towards previously unused habitats, thus also increasing the chances of new disease outbreaks in previously unaffected areas (Woodroffe et al., 2006).

The eradication strategies combining both hunting and carcass removal were the only ones providing at the same time high eradication probabilities within a short time frame (1–3 years) and realistic effort requirements. These corresponded to a 30% yearly hunting rate and to a 30% removal of all infected carcasses, which corresponded to about nine detected carcasses / 100 km2 (Table 2). One of the main benefits of the combined use of both eradication tools was the substantial increase in the probability to remove the ASF virus from the population already during the first year (Fig. 3). Given the stochastic nature of human-mediated ASF introductions into new areas and into the pork meat production system, time to eradication should be considered as a crucial variable when assessing the expected performance of a designed eradication program, because longer endemic persistence times provide higher chances for such stochastic events to occur, eventually causing a geographic expansion or even a long-distance spread (Dellicour et al., 2020, Sauter-louis et al., 2021b).

Overall, given the large difference between the number of infected carcasses currently detected by national surveillance programs (Nurmoja et al., 2017, Oļševskis et al., 2020) and the number needed to reach satisfactory eradication probabilities (Table 2), it is crucial to try and maximise the epidemiological effects of each infected carcass which is found and removed from the environment. To this aim, the analysis of the temporal distribution in ASF extinction events put in evidence that the effort produced to detect carcasses in summer months is unlikely to generate relevant epidemiological consequences (Fig. 5). After spring recruitment, in fact, a new pool of susceptible individuals, the piglets, enters the wild boar population in a short period, suddenly increasing transmission rates and making it almost impossible to eradicate the disease during this time of the year (Fig. 5). Our simulated scenarios (Table 3) suggest that concentrating the carcass removal effort in a shorter period, just before the reproductive season, might provide high eradication chances while requiring the removal of less than 5 infected carcasses / 100 km2, thus substantially reducing the overall field effort (Table 3).

It should be noted, though, that even the most parsimonious scenario, in terms of carcass removal rates, corresponds to a drastic improvement with respect to the current statistics of ASF passive surveillance (Boklund et al., 2018). National and European competent authorities should promote a drastic increase in the effort dedicated to systematically identify and remove as many infected wild boar carcasses as possible from the affected areas. Our modelling results clearly show that, without such a drastic increase, ASF eradication chances will remain low in vast areas of the continent. Given the difficulties in producing absolute demographic and epidemiological data for most wild boar populations, we suggest that a ratio between the number of infected carcasses removed and the number of hunted wild boar could be used as an effective index of the relative effort dedicated by each wild boar managing unit to each of the two eradication strategies. The results of our simulations show that the optimal value for such index varies with wild boar density and ASF prevalence, but that it should be of at least 5–15 carcasses removed from the environment for each 100 hunted wild boar (Fig. 6).

Finally, it should be noted that our modelling results refer to a geographically closed wild boar population, with no immigration and no emigration. Such closure does not allow to consider the probability of new ASF infections coming from the outside, nor the risk that the disease front might expand during the eradication program. While the indications emerging from our study are to be considered valid inside each wild boar management unit, a meta-population approach should be adopted when planning and evaluating nation-wide eradication programs. Further studies are needed to estimate and simulate ASF local colonization and extinction probabilities in different landscape patches, thus allowing to explore the ASF epidemiological dynamics at a larger spatial scale. The other main model assumption, potentially affecting local persistence rates, is that in our model we only considered boar-to-boar transmission rates, either between live individuals or between live individuals and carcasses. In reality, human-mediated disease spread is a potentially relevant, although poorly quantified factor in ASF transmission (Guberti et al., 2019). Human activities which take place in the forest, such as hunting, forest logging, wild boar supplemental feeding, mushroom/berry picking, etc., can play a role in virus displacement and transmission, either because of poor biosecurity measures or because of unintentional environmental contamination of fomites (Mazur-Panasiuk et al., 2019). Such additional role of human-mediated ASF transmission should be further explored, and taken into consideration when defining the limitations and regulation of human forest use in ASF endemic areas.

Funding statement

This work was supported by the European Union's Horizon 2020 Research and Innovation Programme under grant agreement No 773701 (DEFEND).

Ethical statement

The authors confirm that the ethical policies of the journal, as noted on the journal’s author guidelines page, have been adhered to. No ethical approval was required as this is a modelling article with no original research data.

Conflict of interest statement

The authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript.