Chalmers Conferences, 9th European Conference on Mathematical and Theoretical Biology

Modeling of the spread and control of Mycobacterium avium subsp. paratuberculosis in a metapopulation of cattle herds
Gael Beaunée, Elisabeta Vergu, Pauline Ezanno

Last modified: 2014-03-31


The movements of livestock form complex networks linking farms, geographically close or distant, and represent one of the major mechanisms by which many infectious agents propagate at the inter-herd level. Such a pathogen is Mycobacterium avium subsp. Paratuberculosis (Map), which causes paratuberculosis, a worldwide disease that can be considered as mainly introduced into farms by purchasing infected stock. This disease has a significant economic impact on cattle producers, as related to production losses and early culling of clinically affected animals. In addition to animal trade, the diversity of the cattle farming systems in a region (dairy vs beef, variable herd size and structure) and of the contacts among farms (frequency, types of animals purchased) also can influence Map spread and the efficiency of control strategies at the scale of the metapopulation of herds.

Due to the long incubation period and the low sensitivity of available diagnostic tests during the early stage of the disease, studying the infection dynamics in the field is nearly impossible. Therefore, modeling is used to better understand Map spread. Our objective was to study Map spread between cattle herds in order to evaluate, at a regional scale, control strategies based on the management of animal movements between herds depending on their epidemiological status.

Regional infection dynamics was represented by coupling intra and inter herd dynamics. For each herd, a stochastic compartmental model in discrete time described realistic population dynamics and Map spread accounting for all the transmission routes and detailed infection progression. Intra-herd dynamics were then coupled within a metapopulation model through animal movements. Animal trade data from the French cattle identification database (2005–2009) were used. We considered a subset of the network of dairy farms in the Finistère department in Northwestern France, selected with respect to their type and size, this department being characterized by a high density of dairy cattle herds.Various tests at purchase were implemented, defined by the test sensitivity and specificity, the animal age and status at purchase, and the delay between the test and the removal of detected animals (no delay if testing occurred before animal introduction).

The initial number of infected herds, their prevalence, and the network structure highly influenced the speed of Map spread and the endemic state. Testing purchased animals when the test sensitivity is greater than 0.75 prevented the increase in the proportion of infected herds without neither reducing the regional prevalence nor diminishing the within-herd fraction of infected animals. Therefore, a combination of measures, such as test-and-cull and hygiene measures should now be evaluated with respect to the decrease in the within-herd contamination level. In parallel, an extension of the model to include beef herds is under way as animal movements between  this type of herds and dairy herds may also occur.


control strategies; animal movements; modeling of pathogen spread