Global disease suitability models are essential tools to inform surveillance systems and enable early detection. We present the first global suitability model of highly pathogenic avian influenza (HPAI) H5N1 and demonstrate that reliable predictions can be obtained at global scale. Best predictions are obtained using spatial predictor variables describing host distributions, rather than land use or eco-climatic spatial predictor variables, with a strong association with domestic duck and extensively raised chicken densities. Our results also support a more systematic use of spatial cross-validation in large-scale disease suitability modelling compared to standard random cross-validation that can lead to unreliable measure of extrapolation accuracy. A global suitability model of the H5 clade 2.3.4.4 viruses, a group of viruses that recently spread extensively in Asia and the US, shows in comparison a lower spatial extrapolation capacity than the HPAI H5N1 models, with a stronger association with intensively raised chicken densities and anthropogenic factors.DOI: http://dx.doi.org/10.7554/eLife.19571.001

Poultry production is an important way of enhancing the livelihoods of rural populations, especially in low- and middle-income countries (LMICs). As poultry production in LMICs remains dominated by backyard systems with low inputs and low outputs, considerable yield gaps exist. Intensification can increase poultry productivity, production and income. This process is relatively recent in LMICs compared to high-income countries. The management practices and the constraints faced by smallholders trying to scale-up their production, in the early stages of intensification, are poorly understood and described. We thus investigated the features of the small-scale commercial chicken sector in a rural area distant from major production centres. We surveyed 111 commercial chicken farms in Kenya in 2016. We targeted farms that sell the majority of their production, owning at least 50 chickens, partly or wholly confined and provided with feeds. We developed a typology of semi-intensive farms. Farms were found mainly to raise dual-purpose chickens of local and improved breeds, in association with crops and were not specialized in any single product or market. We identified four types of semi-intensive farms that were characterized based on two groups of variables related to intensification and accessibility: (i) remote, small-scale old farms, with small flocks, growing a lot of their own feed; (ii) medium-scale, old farms with a larger flock and well located in relation to markets and (iii) large-scale recently established farms, with large flocks, (iii-a) well located and buying chicks from third-party providers and (iii-b) remotely located and hatching their own chicks. The semi-intensive farms we surveyed were highly heterogeneous in terms of size, age, accessibility, management, opportunities and challenges. Farm location affects market access and influences the opportunities available to farmers, resulting in further diversity in farm profiles. The future of these semi-intensive farms could be compromised by several factors, including the competition with large-scale intensive farmers and with importations. Our study suggests that intensification trajectories in rural areas of LMICs are potentially complex, diverse and non-linear. A better understanding of intensification trajectories should, however, be based on longitudinal data. This could, in turn, help designing interventions to support small-scale farmers.

In recent decades, intensification of animal production has been occurring rapidly in transition economies to meet the growing demands of increasingly urban populations. This comes with significant environmental, health and social impacts. To assess these impacts, detailed maps of livestock distributions have been developed by downscaling census data at the pixel level (10km or 1km), providing estimates of the density of animals in each pixel. However, these data remain at fairly coarse scale and many epidemiological or environmental science applications would make better use of data where the distribution and size of farms are predicted rather than the number of animals per pixel. Based on detailed 2010 census data, we investigated the spatial point pattern distribution of extensive and intensive chicken farms in Thailand. We parameterized point pattern simulation models for extensive and intensive chicken farms and evaluated these models in different parts of Thailand for their capacity to reproduce the correct level of spatial clustering and the most likely locations of the farm clusters. We found that both the level of clustering and location of clusters could be simulated with reasonable accuracy by our farm distribution models. Furthermore, intensive chicken farms tended to be much more clustered than extensive farms, and their locations less easily predicted using simple spatial factors such as human populations. These point-pattern simulation models could be used to downscale coarse administrative level livestock census data into farm locations. This could be of particular value in countries where farm location data are unavailable.