Predicting the risk of Alaria alata infestation in wild boar on the basis of environmental factors.

Alaria alata is an emerging parasite that poses a potential risk for those consuming game, pork, snails and frogs. One paratenic host of A. alata that is known to play an important role in its spread through its feeding habitats is the wild boar. However, no statistical analysis of the influence of aquatic environments and carnivores on the occurrence of A. alata in wild boars has yet been performed. The present study combines a small-scale analysis based on hunting districts in the Mazowieckie province with a large-scale analysis based on data for all provinces in Poland. We applied various modeling approaches, including logistic regression and a generalized linear model in order to determine the presence, intensity and prevalence of A. alata. We used the Alaria mesocercariae migration technique (AMT) to estimate the risk of A. alata among wild boar in a given hunting district or province. The small-scale analysis found that mesopredators (red fox (Vulpes vulpes)) and racoon dog (Nyctereutes procyinoides) were likely to influence A. alata infestation of wild boar; however, the effect was weak, probably as a result of the large home range size of these animals. The large-scale analysis found that wetlands influence the prevalence of A. alata in wild boar, with the estimated risk increasing in the north of the country; this finding is consistent with other studies. Our findings indicate that the occurrence of A. alata in wild boar requires analysis on many levels, and environmental factors play a key role in risk assessment.

Introduction

Even though with the possibility is very low, Alaria alata, a widespread emerging parasite, may pose a potential risk for human consumers of game, pork, snails and frogs (Möhl et al., 2009; Korpysa-Dzirba et al., 2021). The life cycle of this parasite is complex and includes definitive, intermediate and paratenic hosts. The definitive hosts are carnivores such as canids, felids and mustelids (Wójcik et al., 2001; Takeuchi-Storm et al., 2015). In Europe, the definitive hosts are typically red foxes (Vulpes vulpes), wolves (Canis lupus) and racoon dogs (Nyctereutes procyinoides) (Murphy et al., 2012; Rentería-Solís et al., 2013; Ozoliņa et al., 2018).

The A. alaria fluke produces its eggs (a dispersive parasite form) in the digestive system of the definitive host, and they are then excreted into the environment with the feces. These are consumed by intermediate hosts such as snails, tadpoles and frogs, where they develop into miracidia (Portier et al., 2012; Patrelle et al., 2015; Voelkel et al., 2019; Ozoliņa et al., 2021). In the paratenic host, the parasite does not reach the adult stage, but it can survive for months in the muscle or adipose tissue (Riehn et al., 2013). The parasite can later reinfect the definitive host if the paratenic host is consumed. Several mammal species have been described as paratenic hosts (Shimalov and Shimalov, 2001; Rentería-Solís et al., 2018); of these, the wild boar (Sus scrofa) is known to play an important role in spreading A. alata due to its feeding habitats (Ozoliņa et al., 2020).

In the context of public health, it must be noted that A. alata poses a potential risk to consumers of wild boar meat. Alariosis has been described in humans (McDonald et al., 1994; Kramer et al., 1996); however, the etiological agent in this case was Alaria americana, and not A. alata, which occurs in Europe. Considering the close relationships between these two Alaria species and the difficulties in the diagnosis of alariosis in humans (non-specific symptoms), it should be assumed that A. alata may also be a potential zoonotic agent in Europe, as classified by organizations such as the Swiss Agency for the Environment, Forests and Landscape. In such cases, humans may act as paratenic hosts.

As eating frogs is a rather local dietary habit, the most likely route of transmission to humans in Europe is the consumption of infected wild boar meat (Dollfus and Chabaud, 1953). In Europe, the prevalence of A. alata in wild boar is thought to range from 0.6% (France, magnetic stirrer method) (Porteir et al., 2011) to as much as 44.3% (northeast Poland, mesocercariae migration technique) (Strokowska et al., 2020). A. alata larvae are believed to be capable of effective survival in a refrigerator; therefore, as noted by the French Agency for Food, there is a real risk of human infection as a result of eating meat from A. alata-infected wild boar (Korypsa-Dzirba et al., 2021). The most effective method of avoiding infection would appear to be proper heat treatment, as is the case with Trichinella spp. (Gamble et al., 2019). However, considering the popularity of homemade semi-raw meat products, A. alata should not be neglected as a potential pathogen for humans.

The infection rate among wild boar depends on their exposure to the sources of A. alata. Previous studies have reported it to be correlated with the number of foxes (definitive host) living in the same territory (Möhl et al., 2009) and the presence of amphibians in the area (Ozoliņa et al., 2021). The presence of amphibians can be estimated indirectly based on the occurrence of marshland and water areas in a given region. Previous studies have noted a significant difference in the occurrence of A. alata in snails and frogs between seasons: a prevalence of 30% has been observed among snails and frogs in the autumn, and 100% in the spring (Wójcik et al., 2001).

However, no direct evaluation has been performed of the influence of local aquatic environments and carnivores on the occurrence of A. alata in wild boars in a given area. Such data would be of great importance in identifying areas where monitoring should be increased; it could also be used to outline further research directions and support effective preventive activities. Therefore, the aim of the present manuscript was to determine the influence of environmental factors in predicting the occurrence of A. alata infestation in wild boar.

Material and methods

Sample collection and examination

Samples were collected from 576 hunted wild boar from 14 of the 16 Polish Provinces. Provinces (called also voivodeships) are the highest-level administrative division in Poland. The exact numbers of samples taken from individual provinces are presented in Table 1. The procedure used to collect and transport material is described by Strokowska et al. (2020). Samples of muscles, adipose and connective tissue were tested with the Alaria mesocercariae migration technique (AMT) according to Riehn et al. (2010). The characteristic movement and morphological features of this parasite (the body is clearly divided into two sections, with a wing-like shape at the front) were used to assess its presence in tissues. All samples were tested within a maximum of seven days after material collection.

Table 1

Source of data and number of samples for each province included in the analysis.

Province	Present study data	Bilska-Zając et al. (2021)	Strokowska et al. (2021a)	total
Dolnośląskie	10	108		118
Kujawsko-Pomorskie	33			33
Lubelskie	81	7	500	588
Lubuskie		21		21
Łódzkie	19	19		38
Małopolskie	3	3,126		3,129
Mazowieckie	243	1		244
Opolskie		11		11
Podkarpackie	13	30		43
Podlaskie	12			12
Pomorskie	24	2		26
Śląskie	2	58		60
Świętokrzyskie	2	179		181
Warmińsko-Mazurskie	130			130
Wielkopolskie	2	17		19
Zachodniopomorskie	2	5		7
TOTAL	576	3,584	500	4,660

Data elaboration and statistics

Small-scale analysis

Samples with known locations, i.e., where these wild boars were hunted, were assigned to hunting districts. To determine the small-scale impact of variables on wild boar infestation with A. alata, only samples from hunting districts in the Mazowieckie province were examined. The numbers of the most common mesopredators, namely the red fox (Vulpes vulpes) and raccoon dog (Nyctereutes procyonoides), were obtained from the Polish Hunting Association for each of the hunting districts. This data was obtained for 2017. The density of predators in each hunting district was then calculated based on its area.

For each hunting district, land cover data was also obtained from the Corine Land Cover database (CLC) for 2018 (https://land.copernicus.eu). The type of land cover was determined using Quantum GIS (version 3.4.5), which is open source geographic information system (GIS) software. All cover types within the boundaries of each hunting district were calculated with regard to their percentage. Following this, four cover types were selected for further analysis: areas covered by water (referred to with codes 5.1.1 and 5.1.2 in CLC), wetlands (referred to with code 4.1.1 in CLC), arable land (referred to with code 211 in CLC) and forests of various types (referred to with codes 3.1.1, 3.1.2, and 3.1.3 in CLC). Of these, the first two (areas covered by water and wetlands) were expected to have an impact on A. alata infection, while the last two cover types (arable and forests) dominated in the hunting districts.

The impact of environmental characteristics on the occurrence of A. alata mesocercariae in wild boars was determined using a logistic regression model which included all six known explanatory variables for the hunting districts as covariates: density of red foxes as number of individuals per 10,000 km2 (FOX); density of raccoon dogs as number of individuals per 10,000 km2 (RACOON); percentage of areas covered by water (WATER); percentage of wetlands (WETLANDS); percentage of arable land (ARABLE); percentage of forested areas (FORESTS). All A. alata-infected samples were marked as 1; all negative samples were marked as 0. The explanatory variables were verified based on Pearson's correlation coefficient: all values were lower than |r| = 0.7. The quality of the model was verified according to the percentage of correctly classified cases and AUC (area under the ROC curve).

The impact of similar environmental characteristics on the intensity of A. alata mesocercariae in wild boar was evaluated using a generalized linear model with gamma distribution and the log link function. In this model, the dependent variables were the number of A. alata mesocercarie in wild boar (only infected cases); the explanatory variables were (FOX), (RACOON), (WATER), (WETLANDS), and (FORESTS). The variable for arable land (ARABLE) was not included because it was closely correlated with FORESTS (Pearson's r = −0.809, p = 0.000). Model selection was performed according to Burnham and Anderson (2002), where the model presenting the lowest AIC value was chosen as the best one. Akaike weights (ωi) were calculated for each model, and the sum of Akaike weights (∑ωi) for each variable included in the models was within ΔAIC = 2.

Large-scale analysis

The prevalence of A. alata was evaluated against land cover types and red fox density for all provinces. Raccoon dog density was not included due to lack of data. The prevalence of A. alata in wild boar in a given province was estimated based on data from this study and recent literature data (Table 1). All sources employed similar methods of A. alata detection (Strokowska et al., 2021b), which allowed us to increase the number of samples in a given province and minimize bias due to low sample size. In total, the prevalence A. alata was determined based on three sources, with the final prevalence being calculated as a weighted mean of these sources, based on the number of studied samples. The number of samples used and the sources of data are presented in Table 1. The selected cover types of each province were also determined with regard to the four cover-type percentages outlined above, in a similar way as for the small-scale analysis (for hunting districts). A generalized linear model was used with Tweedie distribution and the identity link function, which presented the best values for overdispersion. Model selection (similar to small-scale analysis for infected cases) was performed according to Burnham and Anderson (2002) using the following explanatory variables: (FOX), (WATER), (WETLANDS), and (FORESTS). RACOON was omitted from the analysis due to a lack of data on raccoon dog density in the provinces for 2017. In addition, ARABLE was omitted because it was highly correlated with FORESTS (Pearson's r = −0.748, p = 0.001). All models were verified (including the null model) with regard to the AICc value (for small sample size); again, the model that presented the lowest AICc value was chosen as the best one. Akaike weights (ωi) were calculated for each model, and the sum of Akaike weights (∑ωi) for each variable included in the models were within 95% confidence intervals (∑ωi = 0.95).

The risk of A. alata prevalence in wild boar was also estimated for given provinces in Poland. The prevalence was compared on the basis of the data from the present study, as well as literature values and the prevalence predicted by the model. When predicting the prevalence, only WETLANDS values that significantly explain the prevalence of A. alata in provinces were used. It was also assumed that the prevalence could not exceed 100%; where the predicted prevalence was higher, the value was lowered to 100%.

Results

Small-scale analysis

The occurrence of A. alata mesocrecarie in wild boar was significantly predicted only by raccoon dog density (RACOON) in each hunting district (Table 2). All other explanatory variables were statistically insignificant (P > 0.05). The B coefficient of RACOON was positive, indicating that the probability of A. alata mesocercarie in wild boar increases at higher raccoon dog densities. The model, however, had weak parameters: only 73% of all cases were correctly classified, and the AUC value was 0.614, indicating the model had low accuracy.

Table 2

The effect of FOX, RACOON, FORESTS, ARABLE, WATER and WETLANDS on the probability of occurrence of A. alata mesocercariae in wild boar in the logistic regression model (B – beta coefficient, SE – standard error, OR – odds ratio, N = 196).

Source	B	SE	Wald Χ2	P	OR
Intercept	−0.333	1.107	4.439	0.035	0.097
FOX	−0.023	0.020	1.133	0.249	0.997
RACOON	0.328	0.157	4.369	0.037	1.388
FORESTS	1.816	1.546	1.379	0.240	6.144
ARABLE	1.872	1.417	1.744	0.187	6.500
WATER	−4.855	9.722	0.249	0.617	0.008
WETLANDS	−70.818	74.034	0.915	0.339	0.000

The highest-ranked generalized linear model of intensity of A. alata in wild boar was statistically significant (χ2 = 5.144, df = 1, p = 0.023) and included only FOX (Table 3). All other variables (RACOON, WATER, WETLANDS and FORESTS) were excluded during the selection procedure. Nevertheless, the difference in the explanatory power of the given variables was low because six models were included within the ΔAIC = 2 of the highest-ranked models, and the ΔAIC equaled 0.51 between the highest-ranked model and the next-highest-ranked model (with FOX and RACOON included). Moreover, the ΔAIC between the highest-ranked model and the null model equaled only 3.14 (Table 4). FOX was present in all models within ΔAIC = 2, thus its sum of Akaike weights was the highest (∑ω = 0.5); RACOON and WETLANDS also demonstrated high Akaike weight sums (∑ω = 0.18 and (∑ω = 0.12).

Table 3

The effect of FOX on the number of A. alata mesocercariae in wild boar in the highest-ranked generalized linear model (B – beta coefficient, SE – standard error, CI – confidence intervals, N = 53).

Source	B	SE	Wald Χ2	P	Lower CI	Upper CI
Intercept	2.063	0.202	114.157	0.000	1.766	2.560
FOX	0.034	0.016	4.505	0.034	0.003	0.065

Table 4

Ranking of the models (including the null model) predicting the number of A. alata in wild boar within ΔAIC=2 (ΔAIC – AIC differences, ω – Akaike weights; Rank – rank of the models based on AIC values). Variables: see methods. Best model in bold.

Model	ΔAIC	ωi	Rank
FOX	0.00	0.15	1
FOX + RACOON	0.51	0.11	2
FOX + RACOON + WETLANDS	1.58	0.07	3
FOX + WATER	1.83	0.06	4
FOX + FOREST	1.83	0.06	5
FOX + WETLANDS	1.97	0.06	6
Null	3.14	0.03	10

Large-scale analysis

The highest-ranked generalized linear model for the prevalence of A. alata in wild boar was statistically significant (χ2 = 10.363, df = 1, p = 0.001) and included only WETLANDS (Table 5). The model presented a clearly higher Akaike weight (ω = 0.57) than the other lower-ranked models (Table 6) and differed from the null model by AICc = 7.29. The estimated prevalence of A. alata increased with the percentage of wetland cover in provinces; however, the model predicted a prevalence of over 100% for the Podlaskie province (Fig. 1).

Table 5

The effect of WETLANDS on the prevalence of A. alata mesocercariae in wild boar in the highest-ranked generalized linear model (B – beta coefficient, SE – standard error, CI – confidence intervals, N = 53).

Source	B	SE	Wald Χ2	P	Lower CI	Upper CI
Intercept	5.817	3.586	2.631	0.105	−1.212	12.846
WETLANDS	5,445.837	1,958.886	7.729	0.005	1,606.492	9,285.182

Table 6

Ranking of the models (including the null model) predicting the prevalence of A. alata in wild boars in provinces within 95% confidence intervals (∑ωi = 0.95) (ΔAIC – AIC differences, ω – Akaike weights; Rank – rank of the models based on AIC values). Variables: see methods. Best model in bold.

Model	ΔAICc	ωi	Rank
WETLANDS	0.00	0.57	1
WETLANDS + FOREST	2.98	0.13	2
WETLANDS + FOX	3.34	0.11	3
WETLANDS + WATER	3.64	0.09	4
WETLANDS + WATER + FOREST	7.09	0.02	5
WETLANDS + FOX + FOREST	7.18	0.02	6
Null	7.29	0.01	7
WATER	7.34	0.01	8

Fig. 1

The trend in prevalence of A. alata in provinces with WETLANDS.

The calculated trends of A. alata prevalence in Poland as a whole were similar to the predicted values; however, differences can be observed in particular provinces. On the basis of literature values and data from the present study, our calculations indicated an increase in the prevalence of A. alata in wild boar in the north and north-east (Fig. 2A). These values ranged from 0% in Opolskie in the south-west to 54.6% in Warmińsko-Mazurskie in the north-east, but higher values were also observed in the Zachodniopomorskie (in the north-west, 42.9%) and Podlaskie (north-east, 41.7%) provinces.

Fig. 2

The prevalence of A. alata in wild boar in provinces in Poland calculated from literature values and data from the present study (A) and predicted by percentage of areas covered by WETLANDS (B) (for detailed information, see: Methods). The figure shows prevalence values for a given province and confidence intervals (lower; upper).

Our model did predict a higher prevalence of A. alata in the north (Fig. 2B), with values ranging from 6.8% in Małopolskie, 7.5–7.7% for Śląskie and Opolskie Provinces in the south, to over 100% in Podlaskie in the northeast. Higher values were also observed in the Warmińsko-Mazurskie (43.3%) and Kujawsko-Pomorskie (30.7%) provinces.

Discussion

As predicted, areas covered with wetlands appear to be of significant importance in determining the occurrence of A. alata in wild boar. Our findings also indicate that mesocarnivore density also had an important influence on A. alata occurrence and intensity in wild boar. While the land cover types showed an effect in the large-scale analysis, mesocarnivores demonstrated a weak effect in the small-scale analysis, i.e., in hunting districts. Although our findings are generally in line with current knowledge, some effects seem to derive from more complex relations.

It is not surprising that wetlands have an effect on the prevalence of A. alata in wild boar as the life cycle of A. alata requires an aquatic environment: miracidia, an invasive form of A. alata, are released from eggs into water, thus the first intermediate hosts are freshwater animals (Möhl et al., 2009). It has previously been proven that an abundance of wetlands is positively correlated with the occurrence of adult forms of this parasite in final hosts (Ramisz and Balicka-Ramisz, 2001). This tendency was also confirmed by Tylkowska et al. (2018) in Poland, where the most foxes that were found to be infected with A. alaria were those living near water reservoirs.

Regarding the presence of A. alata mesocercariae in wild boar, Sailer et al. (2012) reported a higher prevalence in an area of Austria that is located in the backwaters of two rivers than an area to the north of one of these rivers, i.e. with less wetland, as described previously by Pestál (1989). Therefore, the lack of this water-related effect in the present study is puzzling. Although this may seem of minor importance, some regions of Poland, especially the northern part, are characterized by large numbers of lakes and of water resources in general (Górniak and Piekarski, 2002). This could partially correlate with WETLANDS and cause a lack of effect by areas covered by water.

Our present findings do not suggest that wetlands or other cover types demonstrated any effects in the small-scale analysis. This may, on the one hand, result from the rough scale of Corine Land Cover, but it might also be caused by the home range size of animals. The mean area of the studied hunting districts was 5,000 ha, which is generally larger than that of the home range of wild boar, which typically cover several hundred hectares (Sodeikat and Pohlmeyer, 2002). However, this home range is dependent on many factors: for example, males and young individuals may have larger home ranges of over 1,000 ha (Mauget, 1980; Keuling et al., 2008). Home range size is also strongly influenced by aspects of landscape structure, such as the number of forest patches or the degree of elevation (Fattebert et al., 2017). In addition, the home range can extend to 3,500 ha under the influence of hunting (Sodeikat and Pohlmeyer, 2002); this effect could also be strengthened by the dispersion of animals as a result of hunting pressure (Keuling et al., 2008). Furthermore, the wild boar has a large dispersive ability, and individuals can migrate up to several hundred kilometers (Keuling et al., 2010; Jerina et al., 2014). This is an important consideration as, during the sampling period, wild boar were under increased hunting pressure aimed at slowing the spread of African swine fever (Klich et al., 2021). Therefore, it would be reasonable to assume that the home range size of wild boar in Poland was significantly larger during the year of sample collection, and the lack of effect that was observed in the small-scale analysis would not be present in other hunting conditions.

In addition to the land cover effect, mesopredators seem to play a significant role in increasing the intensity of A. alata in wild boar. This was noted in the small-scale analysis, in which, despite having somewhat low explanatory power, red fox and raccoon dog numbers were shown to have an effect. These species are present as definitive hosts in the A. alata cycle (Takeuchi-Storm et al., 2015; Korpysa-Dzirba et al., 2021), and the red fox is considered to be the main definitive host of A. alaria in Europe (Portier et al., 2011). Indeed, a recent study in Poland found a high prevalence of A. alata in red foxes (78.7%) based on intestinal examination (Karamon et al., 2020). A. alata does not yet appear have been described in raccoon dogs in Poland; however, it has been confirmed in neighboring Lithuania, and the data suggests that A. alata may even be found in greater abundances in raccoon dogs than in red foxes (Bružinskaitė-Schmidhalter et al., 2012). As such, it might be recommended to test raccoon dogs in Poland for A. alata, especially considering the observed positive correlation between its presence in raccoon dogs and wild boars.

Our data also indicate a clear upward trend in the threat posed by A. alata in wild boar towards the north of the country. This north-south gradient seems to be in line with the results of a previous study of the adult form of this parasite in foxes in Poland which indicated significant differences in prevalence between the northern regions (93.7% and 96.5%) and the southern regions (15.2% and 24.7%) (Karamon et al., 2018). The predicted prevalence in this study in the southern provinces did not exceed 10%, while all northern provinces showed a prevalence of over 20%, and even over 40% in north-eastern Poland. However, the model predicted a prevalence of over 100% for the Podlaskie province, suggesting the model did not have a good fit (Fig. 1.). Therefore, it should be stated that the large-scale analysis also has limitations resulting from the quality of the data used. The calculated prevalence was characterized in some provinces by a large range of confidence intervals resulting from the small sample size (e.g., for Zachodniopomorskie, the calculated prevalence was 42.9%, but CIs ranged from 6.2 to 79.5; for Podlaskie, the calculated prevalence was 41.7%, but CIs ranged from 13.8 to 69.6). Despite the uncertainty of estimation in our study, a similar north-south trend is also visible in the larger European context, with the percentage of positive results being generally higher in northern European countries than in southern ones. For example, the prevalence of A. alaria in wild boar samples, calculated based on the AMT method, was found to be 1.6%, 6%, 11.5%, 43.9% in Hungary, Austria, Germany and Latvia, respectively (Riehn et al., 2012; Paulsen et al., 2013; Berger and Paulsen, 2014; Ozoliņa et al., 2020). In Poland, the southern provinces presented a comparable calculated prevalence to Austria (6.8–7.7% for Małopolskie, Śląskie and Opolskie); the western provinces presented a similar prevalence to Germany (10.9% for Dolnośląskie and 14.1 for Opolskie); and the north-eastern provinces presented the highest prevalence in Poland, which is comparable to the prevalence in Latvia in some cases (for example, 43.3% in Warmińsko-Mazurskie province and 43.9% in Latvia). This trend is also visible among definitive hosts: the positive rates in red foxes were 78.7% in Poland and 94.8% in Lithuania (Bružinskaitė-Schmidhalter et al., 2012; Karamon et al., 2020), compared to 4.7% and 5.3% in Croatia and Italy, respectively (Rajković-Janje et al., 2002; Fiocchi et al., 2016). This trend is probably connected with the fact that northern countries offer generally better conditions for the complete cycle of A. alata, i.e., higher levels of wetland cover (Schleupner 2007).

It should be considered that areas with a particularly high percentage of infected wild boar may be subject to a local epizootic. Such areas should be prioritized for constant monitoring of A. alata. Particularly high infections of wild boar in a limited area have been described in Ireland (Murphy et al., 2012).

Conclusions

As our findings show, the occurrence of A. alata in wild boar requires multifactorial analysis. Environmental factors play a key role in risk assessment of A. alata in wild boar, therefore they should be taken in account when establishing meat inspection strategies and making possible recommendations to hunters in the face of a potential public threat. We recommend that selection of regions for meat inspection should include areas abundant with water and wetlands as well as those with higher local densities of foxes and raccoon dogs. In Poland, such meat inspections should first target northern provinces. A significant limitation occurs when predicting the risk of A. alata infection in small areas, which is probably the result of the large size of the home range of the studied animals.

Ethics approval and consent to participate

Not applicable.

Availability of data and material

The data supporting the conclusions of this article are included within the article. Raw data are available from the first author.

Funding

Not applicable.

Authors’ contribution

DK: conceived, designed and coordinated the study; DK also carried out the study, the statistical analysis and drafted the manuscript. MN, JW: sample collection and laboratory work, manuscript review and editing. AD: drafted manuscript. ZB: sampling, manuscript review and editing. BP: data gathering and analysis. KA: coordinating the study, manuscript review and editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.