A major concern of organizations responsible for monitoring the well being of pastoralists and their livestock in East Africa is making timely, informed decisions with sufficient lead time to allow policy makers to design strategies to mitigate emerging crises. Traditional indicators of animal well being and associated forage supply have focused on body condition of animals, humam/animal movement patterns, water supply, and broad assessment of forage supply. The problem with these indicators is that of timeliness of observations and limited quantification of emerging conditions.
The Livestock Early Warning System (LEWS) proposed for East Africa involves linkage of several new technologies capable of predicting the current nutritional status of free-ranging animals and predicting the impact of weather on forage supply and crop production among a carefully (spatially) selected set of households. The first tool to be used is the Spatial Characterization Tool (SCT) which allows stratification of large regions into units of similar environments called "effective environments". The regional extent of the target human population is then defined based on known attributes of environments that are occupied, in this case the pastoral regions of East Africa.
At issue is to sample these environments in a manner that establishes modality of the variety of landscape in these effective environments. Households, schools, watering points, stock trails and markets are targeted as potential sampling sites. Access is critical and requires an overlay of the road system to determine what routes could be taken to best sample representative or modal units. For each sampling point, the landscapes are classified into virtual landscapes or typical plant community/crop combinations characteristic of the region. The proportion of plant community/crop and the modal composition of plant species growing in each ecological unit are defined.
Once a sampling point has been defined (georeferenced) and virtual landscape characterized for that area, herd populations are estimated based on the sample of the household herd and known population densities in the region. This informationis augmented through interviews with of local extension and NGO working in the area. This essentially sets the demand from the livestock. Also, date of crop planting for those areas with crops is established in the same manner.
Critical to the process is matching weather informatin with each of the selected sampling sites (households). A World Meterological Organization (WMO) station or official government station is assigned to each of the sampling locations. If not available, there is a recording and reporting mechanism establish in a central location to acquire rainfall information and extrapolate temperature and radiation data from other reporting stations.
Finally, we will establish a systematic sampling route that is run every 30 days for a given zone according to the infrastructure available from governmental (National Agricultural Research System (NARS) or Extension) and non-governmental (NGO) entities. At each sampling date, a fecal sample is collected from a composite of at least 5 animals for each of the livestock species (cattle, sheep, goats). The number of deaths, sales, purchases, gifts or transfers of livestock is acquired from the individual to get some estimate of population changes (demand function)>
The enumerator then forwards the samples to a central processing facility for drying and shipment to aregional near-infrared spectroscopy (NIRS) fecal profiling lab. Initially, this will be in Addis Ababa but eventually labs will be established in each country at a location where it can best serve the project. the results are received at the lab, ground through a cyclone mill and scanned with a FOSS 5000 NIRS machine. The resulting predictions of dietary crude protein and digestible organic matter are entered into a nutritional balance analysis model (NUTBAL). NUTBAL predicts changes in ody weight and condition for each sampling site. The model also allow projects the status of the animal for the next 30 and 60 days (see Figures 1 and 2).
Predicted intake demand of the herds is then passed to a grazingland production model (PHYGROW) where 10-day increment weather data are being fed to the system and runs made of forage production under grazing (see example, Figure 3). The models are then run for the next 60 days with current demand and projected temperature under no rain and high probability events derived from historical weather data. Projects of future forage balance are determined. If crops are located at the sampling oint, then the APEX (Agricultural Policy/Environmental Extender) crop simulation model is run as well to make projections on likely crop yields and potential failures.
Given the that the sampling points are representative of a specific area but need to be extrapolated to other areas of similar effective environments. The known hot spots are mapped and then weather and nutrition data are projected to other areas assumed to have similar effective environments to predict likely outcomes for thsoe areas. Hot spots or emerging areas that are not part of the sampling routes are then investigated by rapid deployment teams to verify if critical conditions are emerging and warrant alert status. For those areas identified as alert status, a reporting mechanism is activated to link the information with existing early warning systems such as FEWS (Famine Early Warning System) and the Climate Prediction Center of USGS (United States Geological Survey) as well as direct reporting to governmental entities responsible for policy making concerning disaster relief in each country.
Maintained by the Characterization Assessment and Applications
Group
Blackland Research Center
Temple, Texas
For comments or suggestions contact remartin@brc.tamus.edu