Introduction

Toxoplasmosis is a widespread parasitic zoonosis, caused by the obligate intracellular protozoan parasite Toxoplasma gondii, which can infect all warm-blooded animals, i.e., mammals and avian species (Dubey 2010a). Felids are the definitive hosts of T. gondii (Montoya and Liesenfeld 2004).

Cats can shed millions of oocysts in their faeces that can survive for months even under outdoor conditions. The oocysts sporulate and become infective within 1–5 days, depending on the climate conditions (Dubey 2009). Birds and mammals, including humans, can be infected either by ingestion of sporulated oocysts from water or unwashed contaminated food that has previously been contaminated with cats’ faeces. Other sources of infection can be the consumption of tissue cysts from raw or undercooked meat (Dubey 2009) or vertical transplacental transmission in mammals (Montoya and Liesenfeld 2004, Sarr et al. 2020).

Chickens, especially free-range chickens, have been hypothesized as important intermediate hosts in the epidemiology of toxoplasmosis. Since chickens have outdoor access, feed from the ground, and are susceptible to infection, this species can act as an indicator of environmental contamination with T. gondii (Dubey 2010b; Moré et al. 2012; Hill and Dubey 2013; Dubey et al. 2015;), and as sentinels in regions with a high prevalence of T. gondii infection in humans (Dubey et al. 2002). Therefore, free-range chickens have been widely used to study the prevalence and genetic variation of T. gondii worldwide (Chikweto et al. 2011; Wang et al. 2013; Verma et al. 2015; Vieira et al. 2018a, b; Rodrigues et al. 2019).

A number of studies suggest that poultry may represent an important source of infection for definitive hosts, like domestic cats but also for humans through the consumption of undercooked infected chicken meat (Tenter et al. 2000; Dubey 2010a, b; Stelzer et al. 2019).

T. gondii presents a complex population structure, with three clonal lineages, type I, II, and III prevailing in Europe and North America (Howe and Sibley 1995; Fernández-Escobar et al. 2022). Globally, a number of studies revealed a much greater genetic diversity of the parasite (Khan et al. 2006, 2007; Lehmann et al. 2006; Pena et al. 2008; Su et al. 2012; Lorenzi et al. 2016), subdivided by geographical region and by the existence of non-clonal lineages in areas such as South America, Africa and Asia (Shwab et al. 2014; Galal et al. 2019).

Due to the significant consumption of poultry meat globally, and the potential risk of human infection via this route, elucidating the determinants of spread and risk factors for T. gondii infection in poultry is important. In order to monitor the public health risks and the production of safe poultry products in Greece, a seroepidemiological survey of different types of poultry (backyard, organic, and conventional) was conducted to assess the prevalence of T. gondii infections and identify potential risk factors for T. gondii infection in chickens. T. gondii strains were isolated from the tissues and genotyped, and the co-existence of T. gondii infections with Eimeria spp. was assessed, to elucidate epidemiological links between these two protozoan pathogens.

Material and methods

Study design and sampling

The selection of poultry operations was based on the number of commercial farms in three major Greek regions. Poultry farms in Greece are most concentrated in the geographical regions of Epirus (in North-Western Greece), Central Macedonia, and Central Greece-Attica (Hellenic Ministry of Rural Development and Food, http://www.minagric.gr/index.php/en/). For the purpose of the study, sampling from both commercial operations (slow-growing broiler and layer flocks) and backyard farms was conducted proportionately to their frequency and based on sampling convenience (Fig. 1). The study included 42 poultry operations in Greece, sampled between January 2016 and March 2017. Eight of the operations were raising broilers, 14 were backyard farms, and 20 were layer operations. Layer operations were housing chickens either conventionally caged (n = 8), conventionally floor housed (n = 2), free range (n = 2), or as organic flocks (n = 8). Backyard operations included housed layer hens of different ages. Broiler flocks sampled for this study were slow-growing chicken or commercial broiler breeds kept under conventional (n = 5), free-range (n = 2), or organic (n = 1) production conditions, based on farm availability. In these flocks, the slaughter age had a 75–100 days range. No intensively-raised broilers were included in this study, as earlier studies have shown very low or even zero prevalence (Dubey et al. 2010b; Matsuo et al. 2014; Rodrigues et al. 2019). In the included broiler flocks, animals were sampled at slaughter, when blood samples and the heart were collected for T. gondii detection. In backyard flocks, blood samples were taken from the older animals and in layers from individual animals (at the age of approximately 10 months) according to the expected prevalence. At the time of initial blood sampling, all layer and backyard hens sampled were marked with numbered leg rings on both legs. Depending on producer compliance, flocks were followed up to slaughter where possible in order to re-sample blood and obtain heart tissue during slaughtering. Blood samples (initial and at slaughter) were collected in EDTA tubes, and hearts were placed in individually labelled plastic zipper bags.

Fig. 1
figure 1

Map of Greece reporting the geolocation of organic (blue triangle), backyard (red circle), and conventional (yellow square) tested farms

Full size image

At the time of initial blood sampling, faecal samples were also collected from the ground or litter to be tested in a separate investigation for the presence, quantification—expressed in oocysts per gram (OPG) with a modified McMaster technique—and molecular identification of Eimeria species by multiple PCRs, as described by Andreopoulou et al. (2022).

In total, 1021 blood samples and 322 hearts were collected from the poultry operations (Table 1). All the blood samples and hearts were transferred immediately to the Parasitology laboratory at the Veterinary Research Institute – Hellenic Agricultural Organization Demeter (Thessaloniki, Greece), where they were stored for further examination. The blood samples were centrifuged to obtain serum and hearts were divided into two sub-samples in order to be processed, as outlined below. They were subject to magnetic capture real-time PCR (mc-PCR) as well as for bioassay in IFNɣ-knockout mice to characterize T. gondii isolates. One sub-sample of the hearts and sera were shipped cooled to the Institute of Epidemiology, Friedrich-Loeffler Institute (Greifswald-Insel Riems, Germany) immediately after slaughter, for mouse-bioassay and serologic antibody detection, while the second sub-sample of each heart was shipped to the Institute of Parasitology, Leipzig University (Leipzig, Germany) for T. gondii specific mc-PCR. All samples except the heart tissue for bioassay were stored at − 20° C until further analysis.

Table 1 Study design and samples collected in total
Full size table

In the absence of earlier T. gondii prevalence data from chickens in Greece, a literature-based estimated prevalence of 10% and 5% was used to calculate the sample size per farm, for backyard, free-range, organic, and conventional flocks, respectively. Precision was set at 10% and confidence level at 95%, and the total number of blood samples collected per flock was calculated based on the flock size using Sampsize (http://sampsize.sourceforge.net/iface) and OpenEpi (version 3.01, http://www.openepi.com/SampleSize/SSPropor.htm).

Data collection and statistical analysis

Furthermore, a questionnaire was obtained from all included operations in order to acquire additional information for each farm used for the analysis of risk factors associated with T. gondii infection. Collected farm-specific information included flock size, farm management and biosecurity, nutrition, outdoor area, presence of cats, rodents, or other animals on the farm premises, production rate and performance, disease history, and poultry health (Supplemental Table 1).

For the identification of potential risk factors, a multilevel modelling (generalized linear mixed modelling fit by maximum likelihood (Laplace approximation)) was performed using R (https://www.r-project.org) version 4.0.2, by applying the package lme4. Because individual animals clustered in farms, “FARMID” was included as a random effect variable. The Akaike information criterion (AIC) was used to characterize the relative model quality, as previously described (Lücht et al. 2019; Basso et al. 2020, Deksne et al. 2022).

After multilevel modelling using the individual explanatory variables (univariable multilevel modelling), a final multivariable multilevel model using more than one explanatory variable was established. Starting from those variables with the lowest AIC in modelling, we added further variables with the lowest AIC during the initial univariable testing. The multivariable multilevel model with the lowest AIC was reported as a final multivariable multilevel model. In all statistical analyses, statistical significance was assumed for p values < 0.05.

TgSAG1 ELISA for chickens

Affinity-purified 30 kDa T. gondii surface antigen (TgSAG1, SRS29B; Hosseininejad et al. 2009; Maksimov et al. 2011) was diluted in bicarbonate buffer (0.1 M, pH 8.3) and diluted antigen (30 ng/ml) used to sensitise ELISA plates (120 µl/well, 1 h, 37 °C, PolySorp, Nunc, ThermoFisher Scientific, Denmark). Subsequently, the plates were washed with PBS supplemented with 0.05% (v/v) Tween® 20 (Serva, Heidelberg, Germany) (PBST). A blocking step with 1% casein in PBST (CasPBST; 30 min, 37 °C) followed. The plate was emptied and 100 μl of each serum sample, 1:200 diluted in CasPBST, was added (30 min, 37 °C). After serum incubation, the plates were washed with PBST. Species-specific conjugates (goat anti-chicken IgG (H&L) peroxidase (POD), synonymous to anti-chicken IgY (H&L) (Rockland Immunochemicals; rabbit anti-mouse IgG + IgM (H&L) POD; Dianova, Hamburg, Germany), were diluted 1:4000 in CasPBST. One hundred microliters of diluted conjugate was added to each well and incubated (30 min, 37 °C). Subsequently, the plates were washed with PBST and twice with distilled water. One hundred and fifty µl of 1% tetra-methyl-benzidine (TMB) with 0.012% (v/ v) H2O2 were added to each well. After 15 min at 37 °C, the reaction was stopped by the addition of 50 µl of 2 M H2SO4 and the O.D. in each well read at 450 nm. Each sample was tested in duplicate. Positive (PC) and negative control (NC) experimental chicken sera (Maksimov and Schares, unpublished data) or serum of an experimentally infected mouse, and a pre-infection serum (Schares, unpublished data), were tested in quadruplicate on each plate. To normalise ELISA results, ELISA index values (I) were calculated for each sample (S) based on the means of O.D. values: IS = (O.D.S — O.D.NC)/(O. D.PC O.D.NC). A cut-off optimized for maximum diagnostic specificity was applied (ELISA index 0.242) as previously described for the TgSAG1-ELISASH (Schares et al. 2017).

Mouse bioassay

Mouse experiments (bioassays) reported in this publication were approved by the Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei of the German Federal State of Mecklenburg-Western Pomerania (Permit 7221.3–2.5–001/13).

For the bioassay, IFNɣ-knockout mice (GKO, IFNɣ -/-, C.129S7(B6)-Ifngtm1Ts/J; animal facility of the FLI, breeding pair purchased from The Jackson Laboratory, USA) were used. The IFNɣ response is essential for resistance against T. gondii infection (Schluter et al. 2014). Immunologically impaired mouse strains were therefore used to increase the sensitivity of the bioassay and thus the chance of in vitro isolation of T. gondii from mouse tissue. Heart musculature was pepsin-digested, and one mouse was inoculated per hen. Pepsin digestion was performed as described previously (Dubey 1998; Moré et al. 2012). For comparable results, the volumes of the reagents used for digestion, and the final volume after digestion, were strictly based on the weight of the sample. Briefly, per 5 g of ground musculature (without fat and connective tissue), 25 ml of freshly prepared pepsin digestion solution was added (2.6 g of pepsin (1:10,000 biological activity, Sigma Chemical, St. Luis, MO, USA), 5 g of NaCl, 500 ml of sterile distilled water, 7 ml of HCl (25% v/v) to reach a pH 1.1–1.2). Digestion was performed at 37 °C for 1 h, the suspension was filtered through two layers of gauze, and the filtrate was centrifuged at 1200 g for 10 min. The pellet (originating from the ground musculature) was re-suspended in 2 ml of PBS, neutralized by 1.5 ml of 1.2% sodium bicarbonate (pH 8.3), centrifuged at 1200 g for 10 min and the sediment was re-suspended in 1 ml of DMEM that contained 100 IU of penicillin and 100 µg of streptomycin per ml (all volumes given are per 5 g of musculature, initially pepsin-digested). A final volume of 0.5 ml was used for s.c. injection per mouse in the bioassay. Mice were monitored for 42–47 days. If a mouse developed signs of toxoplasmosis (ruffled hair, apathy) during the observation period it was euthanized according to the Federation for Laboratory Animal Science Associations (FELASA) and German regulations, and necropsied.

For strain isolation, the pleural cavity was flushed with 1 ml cell culture medium as well as homogenized brains and lungs of positively tested mice were inoculated on MARC-145 cells as previously described by Schares et al. (2017).

DNA extraction from pepsin-digested tissues and mouse brains

To extract DNA from mouse tissue and cell cultures, 25 mg aliquots of homogenised mouse brain or 25 µl of cell culture pellets were extracted using the Nucleospin Tissue Kit (Macherey–Nagel, Düren, Germany). To extract DNA from pepsin-digested material, a 200 µl aliquot of the digest was treated with proteinase K as recommended for the Nucleospin Tissue Kit, by scaling up eight times the volumes used for the initial digestion (Macherey–Nagel). After digestion (56 °C, 3 h), 230 µl of the final suspension (1840 µl) was taken and the protocol of the Nucleospin Tissue kit (Macherey–Nagel) was followed as recommended. Approximately, a negative extraction control was represented every tenth sample, which was also tested by real-time PCR identically to the analysed chicken samples.

Real-time PCR

DNA was analysed by a previously published real-time PCR targeting the 529 bp repeat of T. gondii (Talabani et al. 2009), using conditions described previously (Schares et al. 2011; Legnani et al. 2016). A final reaction volume of 20 μl was applied, using a commercial master mix (iQ supermix, Bio-Rad Laboratories GmbH, Munich, Germany) and a CFX96 instrument (Bio-Rad Laboratories). Primers and probes were purchased from MWG-Biotech (Ebersberg, Germany). Real-time PCR primers (800 nM) and a probe (200 nM) were employed as reported (Talabani et al. 2009; Legnani et al. 2016). The cycling conditions were 50 °C for 2 min, followed by 95 °C for 10 min and then 55 amplification cycles of 95 °C for 15 s and 60 °C for 1 min as described (Talabani et al. 2009). After each cycle, the light emission by the fluorophore was measured. Real-time PCR results were analyzed using the CFX manager software Version 1.6 (Bio-Rad Laboratories).

Genotyping of T. gondii isolates

Two methods of T. gondii genotyping were applied, one based on PCR–RFLP and the other based on microsatellites. Genotyping of T. gondii by PCR–RFLP was performed using eight independent, unlinked chromosomal genetic markers (nSAG2, SAG3, BTUB, GRA6, c22-8, c29-2, L358, PK1) as previously reported (Herrmann et al. 2013) by amplifying DNA extracted from cell culture-derived tachyzoites using internal primers as previously reported (Su et al. 2006).

For microsatellite typing, DNAs extracted from in-vitro isolated tachyzoites were genotyped after multiplex PCR, using 15 markers as described (Ajzenberg et al. 2010). These markers included 8 typing markers (TUB2, W35, TgM-A, B18, B17, M33, IV.1, XI.1) that show little or no variation within lineages and 7 “fingerprinting” markers (M48, M102, N83, N82, AA, N61, N60) that display a high level of polymorphism within clonal lineages type I, type II, or type III (Ajzenberg et al. 2010). The only divergence from the original methods was that in the case of M102, AA, and N60, the fluorophore TAMRA was used instead of NED to label amplicons during multiplex PCR.

Magnetic-capturePCR (mc-PCR)

The mc-PCR was essentially performed as earlier published (Opsteegh et al. 2010; Schares et al. 2018) with some slight modifications as outlined below.

Preparation of crude DNA extract

Half of each heart collected (0.9–5 g) was cut using a laboratory grinder or sterile single-use scalpels and then homogenized with the addition of 2.5 volumes lysis buffer (100 mM Tris HCl pH 8.0, 5 mM EDTA pH 8.0, 200 mM NaCl, 0.2% SDS, and 1.2 U/ml of proteinase K) and manual rigorous shaking for approximately 2 min. Samples were placed for overnight digestion in a rocking water bath (85 rpm) at 55 °C and centrifuged for 45 min at 3500 g the next day.

Removal of free biotin

Up to 12.0 ml of heart supernatant was incubated at 100 °C for 10 min to inactivate proteinase K. Fifty microliters of triple-washed streptavidin sepharose was added per sample (binding capacity 300 nmol/ml; GE Healthcare, VWR, Germany). Crude extract samples were allowed to cool down below 40 °C followed by streptavidin–biotin binding in a 45-min rotation in an overhead rotator (10 rpm). A new round of centrifugation at 3500 g for 15 min was performed and up to 10 ml of biotin-free supernatant was transferred to clean 15 ml polypropylene tubes.

Sequence-specific magnetic capture and qPCR on 529-bp repeat element

The sequence-specific magnetic capture, as well as the real-time PCR were processed as performed by Schares et al. (2018). In short, biotin-labelled capture oligonucleotides Toxo CapF and Toxo CapR (Opsteegh et al. 2010) were added to each biotin-free supernatant, and after DNA denaturation and hybridization of the capture oligonucleotides with T. gondii DNA, streptavidin beads were added. After incubation, capture oligonucleotides with hybridised T. gondii DNA were isolated by magnet. T. gondii DNA was obtained from the bead suspension. The subsequent probe-based real-time PCR amplification was done using the previously described primer–probe combination Tox-SC forward, Tox-SC reverse, and Tox-TP1 (Schares et al. 2018).

The mc-PCR results were expressed as Ct values. Results with Ct values < 35 were regarded as positive. Results with Ct values > 40 were regarded as negative. If the Ct value ranged between 35 and 40, the respective amplification curves were visually inspected. If they diverged strongly from those of the positive controls, the samples were regarded as negative.

Results

TgSAG1 ELISA on chicken sera

A total number of 934 chickens (backyard, slow-growing broilers, and layers) were serologically examined by the TgSAG1 ELISA. Eighty-eight chickens (9.4%) displayed antibodies against T. gondii. More precisely, 73 out of 177 (41.2%) backyard chickens and 15 out of 534 (2.8%) commercial layers were seropositive. All tested slow-growth broiler chickens were seronegative. Of the 15 seropositive commercial layer hens, three were kept in caged layer operations, two in floor-housed conventional operations, four in free-range conventional operations, and six on organic farms (Table 2). A different level of seroprevalence was observed in the three sampling regions, with Epirus recording the highest number of seropositive birds (41/313, 13%), followed by Central Macedonia (44/458, 9.6%). In the Epirus district, a 100% (7/7) backyard flock-level seroprevalence was recorded, while in Central Macedonia it was 85.7% (6/7) (Table 2). As for layers, 50% (1/2) of organic farms in Epirus had seropositive hens, with only 5% (1/20) seropositive birds in this farm and a very low overall regional seroprevalence of 1% (1/104), as estimated for seropositive birds per total birds sampled in this region. Layer farms’ seroprevalence in Central Greece-Attica was 50% (3/6), with 3–5% of birds per seropositive farm having T. gondii antibodies and overall regional seroprevalence of 1.8% (3/163), while 60% (6/10) farm seroprevalence was recorded in Central Macedonia with 3 to13% οn-farm seropositivity and a total seroprevalence of 4.1% (11/267) for all birds sampled in that region (Table 3).

Table 2 T. gondii seropositive poultry operations and range of proportion of on-farm seropositive birds
Full size table
Table 3 Seropositive birds per type of poultry, geographical region, and production system
Full size table

Bioassay of chicken heart tissue from the selected seropositive hens

A total number of 26 hens were selected for bioassay, based on the ELISA results (ELISA index > 0.242) and the farms’ and chickens’ availability. Eleven out of 26 heart samples were tested positive for the presence of T. gondii DNA. Bioassay was positive for 7 hens (26.9% of all hens sampled and 63.6% of PCR-positive hens, respectively), all originating from four backyard farms. More specifically, 85.7% of bioassay-positive samples came from backyard farms in Epirus and 14.3% from backyard farms in Macedonia. All seven T. gondii isolates belonged to type II, Apico I (ToxoDB#3). Microsatellite typing confirmed that all isolates came from four different farms. All isolates showed farm-specific patterns in microsatellite fingerprinting (Supplemental Table 2), while only within a single farm, identical typing patterns were observed (Supplemental Table 2).

mc-PCR of chicken heart tissue

Due to impaired sample quality during prolonged tissue shipment, extracted DNA was of sufficient quality and quantity for only 20 out of the 322 sampled hearts. PCR batches were consistently featuring Ct values of around 32 for the 5 × 105 tachyzoites standard and negative controls being negative (Ct > 40). Eight out of the 20 samples were positive (40%) with Ct values of 23–28, indicating a high concentration of T. gondii DNA in these samples. Based on that result, a semiquantitative assessment indicates that there was more than 5 × 105 T. gondii equivalents in the eight positive samples. Seven positive samples were derived from backyard farms (87.5%) in Central Macedonia. One mc-PCR positive sample originated from a free-range chicken, while in this sampling group 0% seroprevalence was observed. One hen was consistently tested seropositive, mc-PCR positive, and bioassay positive.

Questionnaire and risk factors analysis

Questionnaire data was obtained from all 42 farms included in the study. Regarding the facility type, 13 farms in total (31%) had concrete floors in the outdoor area, while the remaining 16 (38%) and 13 farms (31%) had grass-soil and bushland, respectively. Nine (56.3%) grass-soil and six (46.2%) bushland outdoor premises belonged to commercial farms. From a management perspective, 22 farms (52%) had drinking lines with nipples, while 20 (48%) farms had cups; however, the latter was mostly seen in backyard farms (13/20, 65%). Out of the 28 commercial farms, 20 (71%) had hanging feeders, while from the remaining eight farms with ground feeders, five (63%) were organic farms. 79% of the commercial farms had automated feeding (22/28). From the 21.4% of the commercial operations that did not, five out of six farms (83.3%) were organic farms. Strikingly, almost all backyard farms (13/14, 92.9%) had ground feeders, with only 42.9% of them partially using ratio in the nutrition of the chickens, relying mainly on free-grazing (57.1%). The presence of cats, application of insecticides or rodenticides, and regular disinfections were reported in 18 (42.9%), 30 (71.4%), and 31 (73.8%) of all farms, respectively.

For the risk factors analysis, no broiler operations were included in the analysis, since they all tested negative for T. gondii throughout. The univariable generalized linear mixed modelling revealed an increased risk of T. gondii-infection for backyard chickens (p < 0.001), while caged layers in conventional systems were not considered a high-risk group. Chickens in flocks of smaller size were found more often T. gondii seropositive than others. The outdoor area seems to potentially determine the risk level for T. gondii infections; contrary to the presence of concrete floor, it was shown that grass soil (p < 0.01) or bushland (p < 0.001) significantly increased the infection risk. The presence of other animals within the farm premises, such as rodents, dogs, or sheep, significantly increased the chickens’ risk for T. gondii seropositivity (p < 0.01). Regular disinfection, application of rodenticides and insecticides, as well as coccidiosis control were putative protective factors (p < 0.001, p < 0.01, and p < 0.01, respectively). The presence of cats alone was not associated with higher T. gondii infection risk. Other management features, such as having cups instead of nipples (p < 0.001) or round jars (p < 0.05) on the ground (p < 0.001) as watering and feeding equipment, respectively, were statistically significant putative risk factors to be considered. (Table 4). Overall, animals on T. gondii seropositive farms were statistically significantly associated with reports about poor flock performance (p < 0.01) and losses in productivity (p < 0.05) (Table 4).

Table 4 Fixed effects in generalized linear mixed models to determine potential risk factors for T. gondii seropositivity in chicken from Greece (univariable multilevel modelling). Data was analysed by bivariable generalized linear mixed modelling including farm identification number (FARMID) as random effects variable in modelling. The Akaike information criterion (AIC) was used to characterize the relative model quality. Only models with statistically significant explanatory variables (p < 0.05) or variables tending to be significant (0.05 ≤ p < 0.1) are displayed
Full size table

In the final multivariable model, free-grazing practices (p < 0.05) and the absence of an automated feeding system (p < 0.001) appeared to be risk factors for increasing seropositivity (Table 5).

Table 5 Fixed effects in a final generalized multivariable linear mixed model to determine potential risk factors for T. gondii seropositivity in chicken from Greece (multivariable multilevel modelling). Data was analyzed by bivariable generalized linear mixed modelling including farm identification number (FARMID) as random effects variable in modelling. The Akaike information criterion (AIC) was used to characterize the relative model quality
Full size table

No statistical significance was revealed for the seroprevalence differences observed among the three sampling regions, although hens reared in Central Greece-Attica appeared to be protected at a p < 0.1 level (Table 4).

Co-existence with Eimeria spp

On a farm level, T. gondii seroprevalence was 54.8% (23/42 farms). Twenty of the T. gondii–positive farms showed Eimeria spp. oocyst excretion in faecal samples (20/23; 87%). Eight of the combined T. gondii and Eimeria spp. positive farms were backyard operations (8/20; 40%), seven were organic and free-range layer farms (7/20; 35%) and only one was a conventional floor layer farm (1/20; 5%). Seven of the 20 double-positive farms (63.6%) were positive for more than one intestinal Eimeria species. The Eimeria spp. observed in T. gondii seropositive farms were E. acervulina (72.7%), followed by E. tenella (54.5%), E. brunetti (45.5%), E. maxima, and E. necatrix (18.2% each), and E. praecox (0.09%). Risk factors analysis revealed no statistical association between the intensity of shedding Eimeria spp. as expressed in OPG and T. gondii seropositivity. However, although statistically significant only at the level of p < 0.1, co-infection with Eimeria spp. at medium intensity (> 100–1000 OPG) was identified as a risk factor (Table 4).

Discussion

The aim of the present study was to investigate T. gondii seroprevalence and genotypes circulating in chickens in Greece, in both commercial and backyard farms. The overall seroprevalence was 9.5%, which is comparable with chicken studies from other countries (Schares et al. 2017; Tagwireyi et al. 2019; Sarr et al. 2020). We observed significant variations in the proportions of seropositive chickens by poultry type (broilers, commercial layers, backyard poultry). Breaking down the seropositivity results, we observed the highest seroprevalence in backyard chickens (41.2% of backyard birds and 92.9% of backyard farms), which is in accordance with what has been reported for backyard and domestic flocks particularly in Germany (47.7%; Schares et al. 2017), but also Mexico (25.5%; Alvarado-Esquivel et al. 2012), Thailand (64%; Chumpolbanchorn et al. 2009), and Pakistan (17.83% for IgM, 8.8% for IgG; Khan et al. 2020). In layers, a seroprevalence level of 2.8% was identified in the present study. Relatively few studies have been conducted for layers with several of them in China, where a higher prevalence of 7.9% was reported by Xu et al. (2012). Free-range and organic layers had 3.2% seroprevalence, which is similar to the 3.7% seroprevalence reported by Schares et al. (2017) for eastern Germany. All sera collected from slow-growing, free-range, or organic broilers were negative in our investigation, which is different from earlier reports by Rodrigues et al. (2019), although the slaughter age was similar and around 90 days. It seems plausible that prolonged contact with the ground or litter as well as access to the outdoor area including free-grazing practices typically seen in organic and free-range chicken farming, increases the likelihood of T. gondii infection. As the demand for slow-growth and/or free-range chicken meat keeps increasing (Ying et al. 2017), special attention should be paid to T. gondii infections to reduce potential transmission to humans from undercooked meat or other raw preparations like chicken sausages or carpaccio where there might not be sufficient inactivation of the parasite (Schares et al. 2018). The majority of published studies referring to free-range chickens include both commercial and hobby/backyard farms, with a range of seroprevalence from 6.6% to 90% (Asgari et al. 2009; Cui et al. 2010; Zhao et al. 2012; Moré et al. 2012; Chumpolbanchorn et al.2013; Hamidinejat et al. 2014; Gebremedhin et al. 2014; Magalhães et al. 2016; Vismarra et al. 2016; Wang et al. 2016; ElFadaly et al. 2017; Rodrigues et al. 2019; Dubey et al. 2020). This wide range may be due to the use of different serological techniques and cut-off values to detect anti-T. gondii antibodies in the chicken sera, including modified agglutination test (MAT), several enzyme-linked immunosorbent assays (ELISA), indirect fluorescent antibody test (IFAT), and indirect haemagglutination (IHA). IHA is usually considered comparatively insensitive (Dubey et al. 2010b) and is linked with a lower reported seroprevalence in the above-mentioned range. In our study, an in-house T. gondii surface antigen (TgSAG1, p30, SRS29B) ELISA was used, as earlier established by Schares et al. (2017) to assess T. gondii seroprevalence in chickens. The use of T. gondii tachyzoite Surface Antigen 1 (TgSAG1) is meant to avoid cross-reactions with other pathogens (Slany et al. 2016; Schares et al. 2018).

Statistically significant seroprevalence differences were not recorded in the three sampling regions at the level of p < 0.05 although chickens from Central Greece-Attica tended to be less frequently affected at a level of p < 0.1. Prevalence in different geographical areas may be attributed to the sample size, farm housing, and production system, but also to climatic conditions and annual rainfall (Tenter et al. 2000; Dubey et al. 2010b; Mose et al. 2016; Sarr et al. 2020; Khan et al. 2020). The latter parameter is linked with increased soil moisture that can allow oocysts to not only sporulate quicker but also survive longer in the ground (Dubey et al. 2010b). In fact, there are recorded differences in the precipitation among Epirus, Central Greece-Attica, and Central Macedonia. Epirus recorded the highest T. gondii seroprevalence (13%), followed by Central Macedonia (9.6%) and Central Greece-Attica (1.8%), with the annual average rain mm being 90.78, 37.04, and 39.43, respectively (http://emy.gr/emy/el/climatology).

For the molecular detection and isolation of T. gondii DNA, only heart tissue was used in our study. Several studies have confirmed that the heart is the preferred tissue allowing the detection of viable T. gondii stages (Schares et al. 2018; Dubey et al. 2020), which is in line with studies indicating a long-term T. gondii DNA persistence in the heart muscle (Geuthner et al. 2014, 2019). Schares et al. (2018) found that slightly fewer samples were positive if tested by mc-PCR compared to bioassay. In our study, only one sample tested positive for both mc-PCR and bioassay. Potentially, this could be related to insufficient DNA extraction in our study that may have led to failure in the detection of parasite DNA; however, some animals tested positive by mc-PCR but not by bioassay. Another factor impacting inconsistencies between T. gondii seropositivity and mc-PCR positivity could be the inhomogeneous distribution of the tissue cysts (Hiob et al. 2017). We used only half of the heart per chicken for mc-PCR, which reduced the detection potential. However, despite the possibility of false-negative results, mc-PCR is considered a highly sensitive method that can detect even low parasitic loads, as earlier shown (Hiob et al. 2017; Schares et al. 2018). This was also confirmed in our case, where mc-PCR revealed a positive free-range chicken sample, whereas ELISA was negative at this time, suggesting that a combination of serological and molecular techniques might allow earlier and better monitoring of T. gondii infections. We also found four mc-PCR-positive chickens that were bioassay negative. This is in accordance with observations reported by Schares et al. (2018). Interestingly, several chickens were serologically positive both at initial and at slaughter sampling. This hints at a potential long-term presence of antibodies after seroconversion, which might be beneficial for routine testing of chickens used as food animals, especially since overall, seropositivity showed a significant correlation with actual T. gondii DNA findings in the chicken heart.

Determination of risk factors associated with T. gondii infections in chickens is a widely studied field. Flock size seems to play an important role in other studies. In our study, however, observations were rather determined by the type of farm, production system, and husbandry, as smaller flocks (usually backyard, organic, free-range) seem to correlate with higher T. gondii seropositivity rates in chickens. Consequently, backyard and commercial organic farming were revealed as risk factors in our study. These results are in accordance with those of Schares et al. (2017), showing a higher frequency of T. gondii infections in small flocks, compared to large free-range layer operations. Cats are considered a very obvious risk factor due to their role as the definitive host of T. gondii and were found to correlate with T. gondii seropositivity in chickens in several studies (Schares et al. 2017; Vieira et al. 2018a, b; Khan et al. 2020; Chaklu et al. 2020). However, our analysis clearly indicates that the presence of cats alone did not determine the risk or level of seropositivity. Similar findings were reported in many studies where no association between the presence of cats and T. gondii seropositivity on a farm level was seen in small ruminants, poultry, and equids (Stelzer et al. 2019). Possible explanations could be either that cats do not easily enter and contaminate commercial farm premises, feed, or water with oocysts or that cats, even if present, may excrete less oocysts if there is a lower prevalence of rodents. This of course depends also on the type of husbandry and management conditions and should be further investigated. On the contrary, other variables, such as the presence of rodents or other animals on the farm (p < 0.01), like dogs or sheep, as well as insufficient management, e.g., absence of regular disinfections (p < 0.001) or lack of pest control (p < 0.01) significantly increase the risk for T. gondii infections, further highlighting their role in the dissemination of the parasite. Schares et al. (2017), Vieira et al. (2018a, b), and Chaklu et al. (2020) confirmed the presence of rodents and other animals as factors increasing the risk of infection in chicken farms.

Although chickens have rarely been reported to develop clinical toxoplasmosis (Dubey et al. 2010b) and studies around the economic impact and significance of T. gondii infections in poultry are scarce, our risk analysis showed a correlation between T. gondii seropositivity and lower reported productivity (p < 0.05) or reduced performance (p < 0.01). This most likely reflects a more general gap in the sanitary management and biosecurity of the T. gondii–positive farms, although more factors might contribute to the impairment of chicken performance. Earlier findings showed that infection of chickens with genotype II T. gondii oocysts alone caused no clinical signs (Dubey et al. 1993), but impact of the infection was increased by the presence of other pathogens or diseases (Dubey et al. 2010b). In our study, we had four T. gondii seropositive farms, one conventional floor layer farm, and three organic farms, that were positive also for Eimeria spp. and reported flock performance below optimal, but more research is needed to check the actual effect of co-existence on flock productivity.

Outdoor access was found to be an important risk factor for T. gondii infections, with bushland (p < 0.001) and grass or soil (p < 0.01) potentially being optimal to maintain the viability of T. gondii oocysts. Outdoor access allows chickens to feed on earthworms or insects, that may accumulate oocysts from the environment (Stelzer et al. 2019). Earlier studies by Guo et al. (2015) and Maksimov et al. (2011) showed lower T. gondii prevalence in chickens kept strictly indoors, which is in line with our findings. The absence of automated feeding (p < 0.01) along with free-grazing (p < 0.05) is a combination commonly observed in organic layers and backyard farms in our study. This could contribute to the high level of seropositivity of such birds and the fact that high antibody titers were observed at both blood sampling points. Although age was not included in our analysis, we observed the highest seroprevalence in backyard farms where we sampled the oldest animals. Similarly, Chaklu et al. (2020) found that adult chickens are twice as likely to be seropositive than younger ones, which also aligns with the findings by Khan et al. (2020).

Natural co-infections of chickens with T. gondii and Eimeria spp. are considered common (Hiob et al. 2017), just as in other animal species (Mason et al. 2015). In our study, E. acervulina and E. tenella were the species most frequently co-infecting T. gondii–positive chickens. Although little knowledge exists on the dynamics and immunological profile of co-infected animals, experimental co-infection with T. gondii and E. tenella was investigated in vivo (Hiob et al. 2017) and in vitro (Zhang et al. 2018). It appeared that E. tenella tended to display increased merogony in Toxoplasma-positive birds, although this was not accompanied by more pronounced macroscopic lesions in the gut (Hiob et al. 2017). No synergy was observed in co-infected birds by Dubey et al. (2020). We did not observe any clear association between Eimeria spp. OPG levels and T. gondii seropositivity, although there was a tendency that a moderate Eimeria spp. OPG level was more often found in T. gondii seropositive hens. This was not a statistically significant finding but is in line with earlier experimental findings of Hiob et al. (2017) who concluded that E. tenella replication might be depressed in chickens orally challenged with T. gondii oocysts, although the asexual reproduction of E. tenella seemed to be boosted in co-infected birds. However, this assumption needs further investigation. On another note, anticoccidials targeting Eimeria spp. might also exhibit anti-Toxoplasma activity to some extent. Regarding potential risk factors, the increased co-existence observed in smaller flocks and in production systems that are less restrictive in terms of access to the outdoor area might be biased, since the same factors impact the T. gondii mono-infection.

In conclusion, this is the first report of Toxoplasma seroprevalence in chickens in Greece, revealing differences per geographical area and production system. Various management practices have been identified as risk factors. Whether there is a mutual interaction in birds co-infected with Eimeria spp. needs further consideration in experimental studies. Verification of T. gondii tissue stages was possible through bioassay and mc-PCR on top of the serological infection confirmation, however, further research is necessary to establish a combined protocol for early, sensitive, and fast detection of T. gondii infections in chickens for public health purposes.