Evaluation of Breed Suitability, Productivity and Profitability of Smallholder Dairy Cattle production System in Kenya

Abstract

Dairy cattle production contributes approximately 4% of Kenya’s GDP and generates income for more than 1.8 million smallholder farmers in Kenya. Development of the sector is thus very crucial. However, most of the data used in the dairy industry in Kenya are based on sampled surveys, projections and estimates. This makes it difficult to draw concrete conclusions about milk production, consumption, and marketing patterns. A starting point would be an evaluation of productivity and profitability of smallholder dairy breed cattle in Kenya. Determination of the current status of production, marketing, economic statuses/trends, breed profitability and prediction of future milk production would be paramount. The status of smallholder dairy cattle production was described using data collected from four highly dairy and four potentially dairy counties in Kenya using a harmonized county based dairy data gathering and profitability tool. The economic analysis was carried out in highly and potentially dairy cattle farms in Nandi and Makueni counties respectively. In both counties, dairy cattle farm gross margins were used as indicator of economic performance. Two categories were considered i) farms practicing mixed farming, and ii) farms practicing dairy farming only. Further, the study sort to determine exogenous variables influencing dairy farms’ economic performance. Gross margins were determined for all farm cash income less cash costs while multivariable regression using Akaike Information Criterion used to determine exogenous variables influencing gross margin levels. More information was obtained to determine the profitability of different dairy genotypes reared in the smallholder systems. A linear regression model was used in this study to examine the impact of breed type on the profitability of dairy farming while the test day milk yields were used in predicting dairy cow milk yield. Accurate prediction of cow’s milk yield is important in genetic evaluation. The Wood’s incomplete gamma function was fitted to individual lactation curves to predict monthly milk yield. The lactation curve parameters were estimated using the Levenberg-Marquardt’s iterative method in the NLIN procedure of SAS. The Root mean squared error (RMSE), the Mean Absolute Error (MAE), the Mean Absolute Percentage Error (MAPE) and the coefficient of determination (R2) were calculated for the various functions and the predicted and recorded milk yields values were obtained. The daily, partial and lactation milk yields for those dairy cows were computed and the results compared with actual. Results for the status of dairy production indicated that most dairy indicators were not significantly different statistically despite the many years of investment by government and development partners in highly dairy counties. Acreage under fodder for silage (p=0.026), milk production per day (p=0.047), milk production per cow per day (p=0.009), milk consumed in the location per day (p=0.042), number of milk shop retailers (p=0.049), number of AI providers and AI services done per month were significantly different (p<0.05; p<0.1) between the two study counties. Average milk production per farm per day was 7.5 litres and the production per cow per day was 4 litres. Out of the milk sold to processors, only 29% was processed per day in the sampled counties. The findings further revealed that dairy enterprises alone have positive albeit minimal gross margins while the typical smallholder mixed farming (dairy and other enterprises together) result in losses both in Nandi and Makueni counties. On average a dairy farm in the two counties made a profit of Kenya Shilling (KES.) 2,848.30 and KES. 880.80 per year respectively. Final models with exogenous variables had low prediction of gross margins, R2<0.30. The descriptive statistics analysed showed that majority of the dairy cows were Friesian (65%) followed by Ayrshire (32.5%), Jersey (2.2%) and lastly Guernsey (0.3%). The overall profitability mean of the dairy cows was loss of KES. 287,528. This implies that profitability was highest for the farmers who kept Ayrshire at a loss of KES. 284,641 and lowest for farmers who kept Guernsey breed at a loss of KES. 352,561. Overall, initial milk yield increased significantly (P<0.05) across parities while the rate of increase to peak yield and rate of decrease after peak yield were similar for all parities. The Wood’s parameters for individual parities and seasons were all statistically different from zero (P<0.01), but similar in all seasons. Overall, MAE, MAPE and RMSE increased with parity and seasons while the accuracy in prediction, R2, did not have any clear pattern. The correlations between initial milk yield and rate of increase from start to peak of lactation was high and negative and significant (P<0.001). The correlation between actual and predicted daily milk yield in all seasons was between 0.96 to 0.99 while those between actual and predicted peak yield ranged from 0.70 to 0.94. When 305-day was predicted from actual peak yield, the correlation was estimated as 0.71 to 0.87. When the prediction was based on predicted peak yield, the correlation ranged from 0.65 to 0.89. Generally, the findings from this study indicated that due high costs of dairy farm inputs and simultaneous involvement in several farm enterprises, farmers in both counties make losses or realize very minimal profits. Policies for farmers to specialize and reduce feed and labour expenses are highly recommended.