4. DISCUSSION
The Ecuadorian network presents small world properties, strong community structure and a steadily growing tendency over the years, contributing to CSF transmission; these characteristics, as well as the spatial and temporal trends, could be considered by the surveillance system to improve their procedures and update the disease control and management policy, enabling the application of targeted or risk-based surveillance. These results provide a framework for future exploration, relevant to the Veterinary service and stakeholders implementing prevention and control strategies for CSF or any other infectious disease.
Although the Ecuadorian pig network is of medium size (83.000 nodes, 396,000 edges), compared to others, such as those in Slovenia (Knific et al., 2020), Germany (Lentz et al., 2016) and Italy (Crescio et al., 2020), network metrics such as graph density, diameter and small-world properties match those observed in these countries. However, compared to studies in Latin America, it is more similar to that of the state of Rio-Grande-do Sul, Brazil (11,800 nodes, 445,000 edges) (Machado et al., 2021) and larger than networks reported in Argentina (6,300 nodes, 22,000 edges) (Baron et al., 2020), Peru (500 nodes, 25,000 edges) (Gomez-Vazquez et al., 2019), and the state of Santa Catarina, Brazil (10,000 nodes, 50,000 edges) (Cespedes et al., 2021). Although each network is completely different, their size shows the importance of this sector for the Ecuadorian economy and the potential hazards for the spread of diseases across the country or even the region.
More than half of the movements (57%) originated from or were destined for markets, network analysis of cattle shares a similar proportion (53%) (Vinueza et al., 2022), apparently markets have a central influence in the organization of trade in both species, which could be related to the Andean marketing structure (Benítez and Sánchez, 2001) and the livestock population dynamics within the economy (Vernon and Keeling, 2009). This emphasises the high risk of markets as potential spreaders of any animal disease and the need to prioritise them in the prevention and control policies, further analysis could be useful to asses the risks of individual markets and the sales held (Robinson and Christley, 2007), also the possibility of monitoring strategies in this facilities.
Backyard producers constituted 89% of the premises in the network, facing major technological constraints and a lack of biosecurity practices, making them a population at higher risk of CSF infection (Martínez-López et al., 2013). Network metrics such as degree could be used to select those at higher risk and implement target surveillance. Parishes with the top degree centrality contributing with 50% of pig movements, could be prioritised by allocating more human and financial resources, focusing movement supervision on the most likely places to encounter disease (Cameron, 2012).
Higher seasonal movement activity was observed from May to August and December; the last could be related to traditional celebrations, also evidenced in Peru and related to Christmas and New-year festivities (Gomez-Vazquez et al., 2019) when pork meat dishes are highly consumed; in addition, lower seasonal movement activity was observed from January to April, and September to November, with a general behaviour that is repeated across the study years. Thus, the veterinary service could prioritise its actions, modulating their activities according to seasonal network activity rather than administrative or historic decisions.
The communities computed could be important to implement the zoning strategy considered in the Ecuadorian CSF eradication project and implemented in the zoning strategy in Colombia (Pineda et al., 2020) (ICA, 2016). The eradication strategy could be divided into steps, focusing on one community at a time, analysing the risks of infection by incoming movements (Grisi-Filho et al., 2013). The veterinary service has a preponderant role in disease management, closing borders between communities or differentiating requirements for intercommunity movements could be valid strategies, without significant effects on trade (Cardenas et al., 2019), rather than the traditional approach of managing only administrative divisions (provinces).
Some communities showed spatially disconnected parishes due to commercial activity, where producers buy piglets in distant markets and sell them at higher prices than local ones after fattening; also irregular movement schemes could be related.
The border provinces of Loja and El Oro contributed a quarter of the origin and destination of national movements to abattoirs; this proportion is higher than mass marketing centres such as Santo Domingo and Cotopaxi and the major consumption centres in Guayaquil (largest Ecuadorian city) and Quito (capital). We could attribute this result to the underestimation of production, also informal trade along the extensive borders (Terán et al., 2004).
Previous molecular findings from 2015 CSF outbreaks, showed a close relation with strains isolated in 2010 in Peru (Garrido Haro et al., 2018); moreover, the largest Peruvian pig trading community extended close to the border with Ecuador, which is reported to be of higher risk of CSF occurrence (Gomez-Vazquez et al., 2019). Further studies that take these variables into account could be necessary to assess the risk of regional disease spread in border areas.
The network of pig movements contributed to the CSF transmission process; supporting the fact that movement networks are more important than spatial proximity in the spread of CSF (VanderWaal et al., 2020), (Lee et al., 2017), (Rosendal et al., 2014). Pig farming is generally practised indoors (Rodríguez-Estévez et al., 2010), feral pigs could mediate transmission, but in Ecuador they are not commonly seen and no records are kept of them to analyse any relationship. Currently, traders and markets provide the best conditions (direct contact) for CSF transmission and should be prioritised in monitoring and control activities, without underestimation of mechanical transmission and irregular movements.
The lack of unique identification codes for premises in the outbreak database, and the lack of geographic coordinates in the premises database was a limitation in this study.