Integrating knowledge on biophysical and socioeconomic potential to map clusters for future milk production in Ethiopia

Despite growing milk demand and imports, market-oriented milk production and formal processing in Ethiopia is limited to areas around Addis Ababa, notwithstanding its competing land use demand. This study assessed biophysical and market potential for developing the dairy sector, characterizing Ethiopian dairy clusters. Biophysical data from geographic information system (GIS) sources and information from key informants were combined in mapping and ranking these clusters on milk production potential. Twenty-four indicators in six major categories were applied for this assessment: feed availability, environmental conditions for dairy cattle, current production status, access to inputs and services, output market access, and production expansion potential. Feed availability (fodder, crop residues, and agro-industrial by-products as well as land availability and affordability) were the main drivers for dairy development, followed by the current production status, mainly driven by number of (improved) dairy cattle and (formal) milk volumes. Dairy clusters close to Addis Ababa had the highest overall scores for development potential, mainly determined by local demand and access to inputs. For dairy sustainable dairy development in Ethiopia, companies seeking long-term opportunities may avoid the Addis Ababa area and develop dairy production and processing in other clusters especially in Amhara and Tigray regions, with good milk production potential but less developed market infrastructure. The combination of biophysical data and key informant knowledge offered key strengths in delivering valuable results within a short time span. It however requires a careful selection of knowledgeable key informants whose expertise cover a broad scope of the dairy value chain. Supplementary Information The online version contains supplementary material available at 10.1007/s11250-021-02695-2.


Land cover
The land cover product we used is derived from data from another satellite (Proba-V) and distributed by Copernicus Global Land Service. 2 The land cover has a 100 m resolution and is based on the "Land Cover Classification System" (LCCS) 3 by FAO, with 23 main classes, e.g. forests, grasslands, croplands, lakes, wetlands and 10 flexible fractional cover layers (proportional estimates of vegetation cover for several land cover types). As mentioned in the previous section we are especially interested in the grassland and cropland land cover classes (or combinations of the above) which are the ones producing fodder crops or by-products for animal feed.

Heat stress vulnerability
During the summer months, heat stress affects dairy cows and other domestic animals in tropical, subtropical and sometimes temperate regions of the world. Heat and humidity create a sub-optimal condition for dairy cows to produce milk. The optimal thermal zone of dairy cows ranges from ca. 0 o C to 22 o C. If it becomes warmer, cows begin to alter their basal metabolism and metabolic rate. Combined with high humidity levels the effect of heat stress increases. The THI in Figure A2ii demonstrates how 1 https://land.copernicus.eu/global/products/dmp 2 https://land.copernicus.eu/global/products/lc 3 http://www.fao.org/3/a-i5232e.pdf Figure A2i: Actual dry matter productivity (kg dry matter production per ha per year) from (semi-) agricultural areas the combined effects of temperature and relative humidity induce heat stress of dairy cows and the severity of heat stress. An index to estimate heat stress was developed combining relative humidity (vapour pressure in kPa) and monthly mean temperature (in o C). This index is called the Temperature Humidity Index (THI). Itis calculated as: Where: T = Temperature in o C, and RH = Relative Humidity in %. 4 Figure A2ii: Stress categories for dairy cows, based on temperature and relative humidity Monthly mean temperature was combined with monthly vapour pressure data. Using these monthly values per km 2 , the mean THI score per month was extracted.. High producing dairy cows begin to decline production at an average THI of 68. When the index ranges from 72 to 79, cows begin to suffer, and milk production drops rapidly. At THI of more than 80 cows become severely stressed and will not produce milk anymore. In many temperate regions of the world where summers are mild and temperature rarely exceeds 30 o C, moderate to severe episodes of heat stress can occur due to high humidity. THI index of 75 or above can occur when temperature is 27 o C, combined with a humidity above 80%. A second step was to aggregate these values per woreda 5 taking the monthly mean THIvalues of all cells in the woreda. Based on the monthly THI scores per woreda, the maximum THI score per season was calculated. A final value per year is calculated as the average value over all four seasons. If the average maximum THI in four seasons >72 this is classified as "Very High THI scores", if THI is > 68 it is classified as "High THI scores"; all other classes are classified as "Good".

Classification of biophysical potential
The three factors above (biomass, heat stress vulnerability and land cover) were merged to generate a classification of the overall biophysical potential in terms of development of the dairy sector (Figure 8).
The factors can be seen as proxy indicators for 1) environmental conditions for cows, 2) feed availability and 3) level of facilities to enable agriculture, hence the availability of local crop residues and byproducts for dairy production.
Classes are: not suitable (no DMP and THI > 72); low DMP and High THI; low DMP, high THI and low agricultural coverage; low DMP, high THI and agricultural coverage > 50%; High DMP, good THI and >50% agricultural coverage; High DMP, good THI and >75% agricultural coverage; very High DMP, good THI and >75% agricultural coverage; and finally non-agricultural land use.

Additional verification variables
Additional supporting variables were also retained as possible means of verification of the dairy potential, namely distance to cities and cattle density. In Figure 6, we have mapped the Food and Agriculture Organisation (FAO) Cattle density 2006. Although these figures refer to total numbers of cattle and not to dairy cows, they were found to be in agreement with the biophysical potential classes.