Study area and sampling
This study was carried out from 2014 to 2019 in the Basilicata region of southern Italy. This region comprises an area of about 10,000 km2, where the provinces of Potenza (40° 38′ N; 15° 48′ E) and Matera (40° 39′ N; 16° 36′ E) are located. The area is characterized by a Mediterranean climate with dry summers and rainfall concentrated between October and March. Precipitation is abundant, about 1200 mm/year [13]. The average temperature in the coldest month (January) is about +8 °C, and the warmest month (August) about +28 °C, with an annual average of +14 °C.
A geographic information system (GIS) of the Basilicata region was constructed using the administrative boundaries at the provincial and municipal levels as data layers. In order to uniformly sample the animals throughout the study area, the region was divided into 100 quadrants, by overlaying a grid of 10 × 10 km. In each quadrant, about 15 small ruminants aged 3–7 years from 8–9 farms were randomly selected, considering the farmer’s availability to collaborate. A total of 1454 animals (1265 sheep and 189 goats) from 824 farms from all 100 quadrants were examined. The geographical coordinates of each sheep and goat farm were obtained according to the farm code of each farm.
Postmortem examination
The animals from the farms selected from all 100 quadrants during the study were transported to an abattoir for slaughter and postmortem inspection. For each animal slaughtered, CE detection was performed by visual inspection, palpation and incision of the heart, kidneys, liver, lungs and spleen. For each positive sheep, the CE cysts were counted and classified into five morphostructural types (unilocular, multiseptate, calcified, caseous and hyperlaminated) in accordance with Conchedda et al. [8].
When cystic lesions were attributable to CE, the animal and the farm to which it belonged were classified as positive.
Molecular analysis
The molecular study was carried out on 353 cysts (300 from sheep and 53 from goats). At random, hydatid fluid (where present) or the parasitic membrane was obtained for molecular analysis [14]. From 300 sheep (179 from liver, 102 from lungs, 11 from spleen, eight from kidneys) and from all 53 goats, one cyst for each animal was collected (independently of the morphotype of cysts).
The cysts and the cystic liquid were collected and stored at –20 °C until DNA extraction. Genomic DNA was extracted using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) [14]. Polymerase chain reaction (PCR) for the cytochrome C oxidase subunit 1 (CO1) gene was performed as reported in Capuano et al. [14], while PCR for the 12S rDNA gene was conducted as described in Rinaldi et al. [15]. PCR products were detected on a 2% ethidium bromide-stained low melting agarose gel (BIO-RAD, Spain) for both PCR reactions. Bands were cut from the gel under UV exposure, and the amplified DNAs were purified using the QIAquick Gel Extraction Kit(Qiagen, Germany). The PCR products were sequenced and analyzed using Chromas version 2.6.6 software. DNA sequences comparison was achieved using GenBank with the BLAST system and ClustalW.
Geostatistical analysis
All georeferencing and data were expressed in geographical ETRS89 format and were projected to UTM zone 33N at reference datum WGS84, as specified by the RSDI [Regional Spatial Data Infrastructure] Basilicata Geoportale [16].
Indicator kriging to access continuous area probability
Disease incidence detection and probability mapping were performed in three steps. The first step produced empirical semi-variograms, which represented half of the mean square difference between pairs of sampling locations (Eq. 1).
$$\gamma \left( h \right) = \frac{1}{2N\left( h \right)}\mathop \sum \limits_{i = 1}^{N\left( h \right)} \left[ {z\left( {x_{i} + h} \right) - z\left( {x_{i} } \right)} \right]^{2} { },$$
(1)
where N(h) is the number of data pairs for the lag h, while h is the distance between animal sampling sites and z(xi) is the location of the animal sample. The stable semi-variogram function [17] was used to fit the semi-variogram model to the empirical data.
The second step involved estimation mapping to predict the presence or absence of disease in an unknown location. Indicator kriging was used to estimate mapping distributions under a given threshold [18]. The resulting data were interpreted as values between zero and one. The greater the value, the more probable the occurrence of the event, i.e., higher probabilities indicate a greater likelihood of finding a farm with an infected animal.
The last step consisted of estimation mapping for the probability of presence or absence in the range 0–1, as described in Adhikary et al. [19].
Local Moran’s I statistics for spatial autocorrelations and clustering
Local spatial autocorrelations were used to calculate the significance levels of local indicators of spatial association (LISA) based on observed farms throughout the study area.
In this study, LISA [20, 21] was used to reflect the degree of correlation between the incidence of disease among animals on a given farm and the incidence among animals on nearby farms. The local Moran’s I index was defined as:
$$I_{j} = \frac{{n\left( {X_{i} - \overline{X}} \right)\mathop \sum \nolimits_{i = 1}^{n} W_{ij} \left( {X_{j} - \overline{X}} \right)}}{{\mathop \sum \nolimits_{i = 1}^{n} \left( {X_{i} - \overline{X}} \right)^{2} }},$$
where n is the number of space units involved in the analysis, Xi and Xj represent the observed values of a phenomenon (or an attribute characteristic) x on the I and j of the space unit, and Wij is the spatial weight generally based on a point distance function.
The Ij value can be mapped to highlight data based on relative importance and surrounding behavioral association, leading to four categories:
-Low-low (LL): a point with a low value with surrounding points with low values (positive Ij = same behavior), interpreted as a “cold spot cluster”;
-High-high (HH): a point with a high value with surrounding points with high values (positive Ij = same behavior), interpreted as a “hot spot cluster”;
-Low-high (LH): a point with a low value with surrounding points with high values (negative Ij = different behavior), interpreted as a “cold outlier spot”;
-High-low (HL): a point with a high value with surrounding points with lower values (negative Ij = different behavior), interpreted as a “hot outlier spot”.
These categories can facilitate a direct interpretation of behavioral phenomena over the entire area.
Based on the spatial location of the LL and HH points, two ellipses were constructed in order to approach its spatial dimensionality, using a distance between farms approach.
All these results were utilized to construct a map for the purpose of interpreting disease behavior in a whole view, providing a “complete picture.”
All analyses were performed using the ESRI ArcGIS ArcMap 10.6 software.