Cross-sectional study of productive and reproductive traits of dairy cattle in smallholder farms in Meru, Kenya

J Muraya, J A VanLeeuwen, G K Gitau¹, J J Wichtel², D N Makau, M B Crane, S L B McKenna and V T Tsuma¹

Department of Health Management, Atlantic Veterinary College, 550 University Avenue, Charlottetown, Prince Edward Island, Canada C1A 4P3
jmuraya@upei.ca
¹ Department of Clinical Studies, Faculty of Veterinary Sciences, University of Nairobi, PO Box 29053-00625, Nairobi, Kenya
² Ontario Veterinary College, University of Guelph, 50 Stone Road E., Guelph, Ontario, Canada, N1G 2W1

Abstract

A cross-sectional study was conducted to determine the farm management and milk production and reproductive performance of dairy cattle in smallholder dairy farms in eastern rural areas of Kenya, and to determine farm- and cow-level factors associated with milk production. A total of 200 farms were randomly selected from a list of the farmers shipping milk to a local dairy society. Structured questionnaires were used for data collection on management and demographic information, and farm visits occurred where the lactating cows on the farm received a physical examination. A mixed linear regression model with a random effect for farm was fit to determine associations with the natural log of daily milk production.

The majority of the farmers had one to three milking cows (mean = 1.40), with an average milk production of 6.70 kg/cow/day from the 314 lactating cows on the 200 farms in the study. At the time of the study, 43.4% of the lactating cows were bred and/or pregnant, with 28.7% of the cows being confirmed to be over three months pregnant. The cows that were cycling and non-pregnant (n=74) had a mean of 304 days-in-milk (DIM), while those cows that were anestrous (n=95) had a mean of 201 DIM.

Explanatory cow- and farm-level variables in the final milk production model were reproductive status of the cow, breed type, weight, DIM, dairy meal fed during the last month of pregnancy and land allocated for growing fodder for dairy cows. Exotic breed crosses, producing 6.80 kg of milk per day, on average, had higher milk production than the indigenous breeds, producing an average of 4.90 kg of milk per day. Heavier animals yielded more milk on the day of the visit; cows that weighed over 550 kilograms had twice as much milk production as those that weighed 250kg and less. The study categorized the cows into different reproductive statuses (early pregnancy/anestrous, pregnant, and cycling) and noted a steady increase in milk produced by cows in these different groups, with the cows that were cycling recording a 19.8% higher daily milk production over those in early pregnancy or anestrous. Milk yield reduced steadily as DIM increased beyond the first hundred days. Milk production from cows that received dairy meal in the last month of gestation was 34.3% higher compared to those that did not receive any. The percentage of land allocated to growing fodder for dairy cows was positively associated with the cow’s milk yield per day, with a 15.6% increase for every 25% increase in land set aside for growing fodder.

We conclude that, even though smallholder dairy farmers in this area of Kenya have made attempts to improve their animals by cross-breeding them with exotic breeds, the milk production was still low. This can likely be largely attributed to poor feeding (especially as young-stock and during the transition period) and reproductive management. A more detailed cohort study or trial is recommended that can examine all the changing cow and management factors over time, providing necessary recommendations for farmers that account for these changes over time.

Key words: days in milk, mixed model, pregnancy, reproductive performance, transrectal palpation

Introduction

The smallholder dairy (SHD) sub-sector in Kenya accounts for 80% of the total number of cattle in the country, contributing to 70% of the total milk output (IFAD 2006; Odero-Waitituh 2017). Irrespective of the large numbers of animals, per cow milk productivity of the dairy sector is still very low. The SHD farmer is faced with limitations to achieving optimum milk production, including poor management, poor nutrition, lack of desirable breeds, infertility, reproduction disorders, animal diseases and a poor marketing system (VanLeeuwen et al 2012). The reproductive performance of the herd or animal is a key indicator of sustainability of a dairy farming system (Swai et al 2007). In the north American dairy sector, if a cow cannot show heat promptly, conceive at an optimal time, and deliver a calf per year, lifetime milk production is suboptimal and the enterprise is not considered very profitable or sustainable (Hare et al 2006).

Assessment of reproductive performance depends on composite parameters, with the main indices being average Calving Interval (CI) and days open. Average days open has been advocated as the most appropriate measure of current reproductive performance (Radostits et al 2001), but for SHD farms, this measure is too variable with the small herd size. In order to achieve the optimal CI of 12-13 months, a Calving-to-Conception Interval (CCI) of 85-110 days is recommended (Radostits et al 2001). These intervals are negatively influenced by biological (postpartum diseases, delayed resumption of heat, and cystic ovarian disease) and management factors (poor nutrition, heat detection problems, poor breeding techniques, and long voluntary wait periods) (Radostits et al 2001).

Reproductive performance in smallholder dairy enterprises in Kenya has been described as poor (Odima et al 1994; Bebe et al 2003; Owen et al 2005) . It is characterized by long calving intervals of about 633 days (Bebe et al 2003). These low reproduction indices, together with high youngstock mortality rates, have resulted in farmers being unable to produce enough replacement heifers. In order to overcome these and many reproductive constraints facing farmers, effective input services are required. Interventions from the government in terms of service provision and subsidies, and also strengthening of farmers’ cooperative societies, are ways of achieving these reproductive goals. Romney et al (2000) reported that in the Kenyan highlands, farmers were willing to purchase supplemental feeds when given access to credit facilities.

Previous studies on reproductive performance in Kenya have been done in the 1990s and early 2000s (Odima et al 1994; Bebe et al 2003; Owen et al 2005), and there have been efforts to improve reproductive performance since those reports. These studies however were done in the peri-urban and urban areas surrounding Nairobi, and management, ecological and production factors in these peri-urban areas are not similar to those of rural settings of Kenya, such as the Meru highlands.

Due to Kenya’s steady population growth, progressive land subdivision has been ongoing and that has rendered these small portions of land too small for subsistence crop agriculture (Asoka et al 2013). Small-scale farmers in these rural areas have now intensified dairy production as their main source of income, and this has opened a door for the need to improve their production and reproduction. For this intensification to happen, studies are needed to determine the state of the industry and the challenges farmers are facing, and to make feasible recommendations that will lead to the improvements needed. This study was therefore designed to determine the production and reproductive performance of dairy cattle in smallholder dairy farms in the Naari area of Meru, and to determine associations between reproductive status (main predictor of interest) and milk production (outcome of interest), while investigating other important variables and controlling for confounding. The results of this study provide a baseline assessment for a larger project that involved a dairy cooperative society, a Canadian non-governmental organization called Farmers Helping Farmers, and supplementation of nutritional and reproduction interventions for their animals.

Methodology

Study area

The study was conducted within 10-15 km radius around the rural area of Naari in Meru County, Kenya. Meru County is located in the eastern parts of Kenya (longitudes 37^o 18’ 37” to 37^o 28’ 33” east and latitudes 00^o 07’ 23” to 00^o 26’ 19” south), approximately 270 km north of Nairobi, and has a population of approximately 1.5 million people, of whom, 84% reside in the rural areas (Mutarari 2010). The precipitation in this county is bimodal, with short rains around the months of March to May and long rains around the months of October to December. The highest amount of rainfall approaches 2200 mm in the highest altitude areas of this county, while only 500 mm may fall in the lowest altitude areas of the county. Average daily temperatures in the highlands range between 14^oC to 17^oC while those of the lowlands are between 22^oC to 27^oC. Agriculture is one of the main economic activities in Meru County, with both cropping and livestock being common. According to the welfare monitoring report by the government of Kenya, the percentage of households living below the poverty line in 2008 in Meru North was 44.7, with this number expected to rise (GoK Meru-North distict Development Plan 2004-2008).

Study farms and cows

The sampling frame for the study consisted of 568 farms that were identified from the Naari Dairy Farmers Cooperative Society (NDFCS or Naari Dairy) database as active members shipping milk to the Dairy in the month of February 2015. A total of 200 farms were randomly selected from the database for the study.

In computing the necessary sample size, a confidence of 95% and power of 80% were assumed using Epi-Info version 6.04b (CDC Atlanta USA 1996) to detect associations between the main dependent variable (milk production) and the main independent variable (reproductive status – pregnant or not), based on a mean milk production of 5.52 kg/cow/day in pregnant cows and 6.69 kg/cow/day in non-pregnant cows, with a combined standard deviation of 1.41 kg/cow/day (Melaku and Gurmessa 2012). Sample size estimation results indicated 200 farms were necessary from the sampling frame of 568 smallholders for the study. Farms were estimated to have 1-3 milking cows (2 cows on average) for this sample size to be considered adequate. All milking cows from the selected farms were included in the study.

Data collection

The farms were visited once (cross-sectional study) during the period of May-August 2015, and a questionnaire was administered to collect all the relevant information. This involved detailed tracing of all milking cows on the farm, and examination of any written records, if any, so that all ages of the cattle, calving dates, history of reproductive diseases and conditions around parturition, such as mastitis cases, were recorded chronologically. Other information collected through the questionnaire included details on feeding and mineral supplementation, whether the cattle owner had attended any dairy husbandry training, herd size, awareness and monitoring of heat signs, age of the cows and source of animals.

The animals were examined physically, and the following information was collected: live weights using a weight measuring tape around the girth area, height at the withers, body condition scored on a 5 point scale where 1 represented very thin and 5 represented grossly overweight, using half point increments (Nicholson and Butterworth 1986), and any clinical abnormalities. A California mastitis test (CMT) was performed, and pregnancy status and ovarian status were obtained by way of transrectal palpation.

Definition of reproduction parameters

Days open was calculated as the period between the last calving and conception if the cow was pregnant, or the visit date if the cow was open. It was hard to get the actual days open for all cows since farmers did not practice good record-keeping, and their recall of dates when the cows were served or last delivered a calf was approximate to the nearest month. Therefore estimates of days open used the 15^th day of the month reported for calving and breeding where there was no physical record of them.

Days in milk (DIM) was defined as the number of days during the current lactation that a cow had been milking, beginning with the last date of calving to the current date. Abortion was defined as the expulsion of one or more calves <271 days after natural mating or artificial insemination. Foetal membranes were considered retained if they remained unexpelled for at least 24 hours after calving or abortion. Dystocia was considered to occur if parturition was assisted either by the farmer or by a veterinary field officer.

Statistical analysis

Data were entered and organized in an Excel spreadsheet (Microsoft, Sacramento, California, USA). The unit of analysis was the individual lactating cow in the farm at the time of visitation. Descriptive statistics for the animal- and farm-level variables and analytical statistics were carried out using STATA/IC 13.0 (StataCorp LLC, College station, Texas, USA).

For the analytical statistical analyses, the main outcome (dependent) variable investigated was the reported natural log of kilograms of milk produced per cow per day for the day prior to the visit. Due to the lack of records kept by most farmers, leading to possible measurement error, continuous variables of age, days open and DIM were modified into categorical variables to minimize information bias for the analytical statistics. Farm was included as a random effect because cows on one farm are not statistically independent of one another (Kristula et al 1992).

In the first step of the modeling, relationships between each independent variable and the outcome variable were individually investigated. In the second step, any variables that were associated at the p<0.15 level were eligible to be included in multivariable models. Correlation matrices between variables meeting the cut-off level (-0.3<r<0.3) were examined to determine correlations among these variables. Both forward stepwise and backward elimination regressions were used to identify the most parsimonious model in which all independent variables retained at the p< 0.05 level. Other variables not in the final model were examined for confounding of the variables in the final model, as recommended (Dohoo et al 2009). Interactions between variables in the final model were investigated. Model fit was examined by checking the standard residual diagnostics, performing predictions, and checking shrinkage of the model used.

Results and discussion

Farm characteristics and management

Since all 200 selected farms agreed to participate in the study, there was a 100% response rate. The principal farmers were primarily women (52.5%), although there were instances where both the male and female jointly considered themselves as principal farmers (16.5%). Most of the principal farmers were married (79.5%), but a few of them were young people who were single and establishing themselves as dairy farmers (9%). The majority of the principal farmers had either none or primary level of education, whether male (56%) or female (57%), indicating the low literacy levels among the farmers, and leaving a huge need for training on dairy farming matters. The mean (+SE) household size recorded in this study was 3.78 ± 0.12 with a minimum of 1 person and a maximum of 11. Higher man’s education was positively associated with log of milk production as an ordinal variable in the univariable regression analyses (p<0.05), with 31.9% and 7.0% of male farmers having completed secondary and tertiary education, respectively (Table 1).

Among the farmers interviewed, 61% of them indicated that other than dairying, they also practiced crop farming, which supported their source of income and food, while the crop residues were used as feed for their cows. Only 13% of the farmers had wage or salaries coming to either them or their spouse, while 10% of the farmers had no other source of income other than the dairy cows. It has been reported that cattle production plays an important role in improving the livelihood for farmers in Kenya (Thornton 2010), and our research would corroborate this assertion. In our study region, cattle were mainly kept for food and cash income (milk and/or meat), but also for draught purposes and farm manure/fertilizer. Source of income met the eligibility criteria for multivariable regression modeling (Table 1), being marginally associated with natural log of milk production as an ordinal variable (p<0.05).

The mean total land holdings owned by the respondents was 2.04 ± 0.17 acres, although the farmers indicated having access to other pieces of land in the form of leasing it, borrowing it, or using part of the nearby government-owned forest which was leased to them for some time. These additional portions of land were small and made up a mean of 0.41 ± 0.06 acres. A majority of the respondents (51%) indicated that, of all the combined land pieces, they allocated between 25-50% of the farmland to growing feed for their dairy cows, and this was because dairy farming was considered the major source of income. Land allocated for dairy use was positively associated with natural log of milk production as an ordinal variable (Table 1) in the univariable regression analyses.

Half (51.2%) of the farmers indicated that they had obtained their milking cows through purchasing them as adult cows, as compared to those who purchased them as youngstock (28.4%) or raised them on their farm (20.4%). The animals were reported to be obtained from the neighbouring smallholdings within the greater Meru County, as buying from large-scale establishments in Rift Valley and Central provinces of Kenya was considered expensive and those animals were less adaptable to the local challenging feeding management. Purchased animals had also been indicated as a common source of cows in a study in the Kenyan highlands nearby (Kiambu, Machakos, Kirinyaga, Maragua, Nakuru, Nyandarua and Narok former districts) for supplying milk to Nairobi (Bebe et al 2003). However, the farmers also indicated a preference for raising heifers born on their farms as replacement stock as it was considered cheaper than purchasing an animal, and the fertility and production history was known for home-raised heifers.

Artificial insemination (AI) services were readily available, and offered by private practitioners, government veterinary officers and veterinary technicians. However, 13% of the study farmers still preferred to use bulls for breeding. Even among the farmers that used AI for breeding their animals, a majority did not have the knowledge to choose which sire to use on their cows, with the majority allowing the AI or veterinary technicians to choose which bull to use, or to advise them on what bull to use, even though most of them had attended some form of dairy training. A few of the farmers were specific in their answers, saying they used “imported” or “Canadian” sexed semen on their cows, and reporting “good results” with that semen as well. This semen is usually highly priced in Kenya, and so most farmers shy away from using it, especially due to reported perceived low conception rates compared to regular semen (Norman et al 2010), potentially leading to a repeat service. Kenya’s Animal Genetic Resource Centre (KAGRIC) is responsible for keeping the AI bulls and distributing semen in the country, as well as in neighbouring countries (Wakhungu et al 2000). There is also a presence of imported gametes, in terms of semen and embryos, that come into the country through the veterinary services office, and lately, sexed semen from different countries has been made available through this avenue (APSK 2015).

The basal dairy cattle feeds in our study were based on natural pastures and home-grown fodder, mainly maize stover, Napier grass and crop residues. Of the 200 farms, 73% zero-grazed exclusively, while the remainder utilized cattle grazing on their land at least some of the time. Napier grass contains moderate crude protein (CP) content (6-12%) when it is fed at 1- 1½ meters in height, but declines to less than 5% when it is fed at 2½ - 3 meters in height (Njoka-Njiru et al 2006). When natural pastures and other cultivated pastures are available during the rainy season, Napier grass is usually not fed to animals but instead is left to grow tall and then fed during the dry season, usually leading to milk production dropping substantially.

During the dry season in our study, maize stover was a common crop fed to cows; over 80% of the farmers reported using it. Although dry maize stovers are important sources of roughage, they have low nutritive values with CP as low as 2.5% of dry matter, and neutral detergent fibres exceeding 70% of dry matter, making them a poor choice for lactating dairy cow feed. Crop residues that were available and sometimes fed to the study cows included relatively nutritious cowpea pods (7.10%), bean pods (63.7%), and sweet potato vines (15.5%). However, poor storage methods practiced by the farmers predisposed the crop residues to rains and sunlight, likely resulting in further deterioration of the nutritive quality of the feed.

To counteract some of this diminishing quality of feeds, concentrates were usually fed to cows, with dairy meal being the principal commercial supplement offered. Milling by-products, such as wheat or rice bran, wheat pollard and maize germ, have also been used as they are seen as a quick cheap source of energy for the cows. All of these products were available in the Naari Dairy consumer shop or at feed stores located in the local shopping centres. Farmers indicated their preferences to using the Naari Dairy consumer shop since they had access to credit there as Naari Dairy members. About 84% of the farmers indicated feeding dairy meal to their lactating cows, while only 58% were giving dairy meal to dry cows during the transition period. The 16% of farmers that did not give dairy meal to their cows cited high cost as the main constraint, and they were occasionally feeding the cheaper milling by-products mentioned. Dairy meal fed to cows on the farm in last month of gestation was positively associated with natural log of milk production as a dichotomous variable in the univariable regression analyses (p<0.05).

Mineral supplements in the form of powdered salts, blocks or molasses were available to farmers in this area. A total of 88% of farmers fed mineral powders to their cows, and 48% of those not giving the powders indicated using mineral blocks. Molasses was only used in the dry season and mixed with dry fodder to increase its palatability.

The quantity of mineral and dairy meal supplements fed was generally low, and in most cases, a fixed amount was used throughout the lactation without adjustments according to the amount of milk produced. For example, 33% of farmers were giving cows a 1 kg can of dairy meal twice per day to lactating cows (equivalent to 1.3 kg of dairy meal per day), regardless of milk production or desire to get the cow pregnant. Similar findings had been reported in the central highlands of Kenya (Rufino et al 2009) and in the semi-arid areas of eastern Kenya (Njarui et al 2011). It was clear that most farmers were unaware that not providing the required amounts of mineral and dairy meal supplementation to lactating dairy cows will lead to lower milk production and delayed conceptions (Moran 2005).

Cow variables

The 200 farms had 316 total milking cows at the time of the study. There were two cows that were very hostile, and therefore a transrectal palpation was not carried out to confirm their reproductive status, although they were reported to be open. Therefore, reproductive status results are based on 314 cows.

Mean and median milk yield of the 316 lactating cows was 6.7 ± 0.23 and 6.0 kg/cow/day, respectively, with 35% of the farms producing less than 5 kg/cow/day at the time of the study, while the upper 10% produced over 12 kg/cow/day of milk, on average. Milk yield was not normally distributed, and 3.8% of the farms produced more than 15 kg/cow/day. As reported elsewhere, this low average milk yield could be attributed to underfeeding of lactating cows and giving poor feed quality, since most of the farms in smallholder dairy farming in Kenya rely on Napier grass as the main roughage, which can be very poor in quality if it is allowed to grow to 2 metres or more (Omondi and Njehia 2014).

Dairy stock kept included Bos taurus crosses (Friesian, Ayrshire, Jersey, and Guernsey) and Bos indicus crosses (Zebu, Boran). A majority (48.1%) of the respondents preferred Friesian to Guernsey (31.9%) or Ayrshire or Jersey crosses (12.3%), due to the perception that Friesian cows have a higher milk production. The Friesian-Holstein crosses produced an average of 7.50 kg of milk per day, which was the highest production of all the breeds. Guernsey crosses gave 6.24 kg/cow/day, which was higher than the Ayrshire or Jersey crosses (5.38 kg/cow/day). The least common breed (Zebu or other indigenous crosses) only produced 4.90 kg of milk per day on average. However, DIM and other factors affecting milk production are not taken into account in these means, and therefore multivariable model coefficients that control for other production confounders should be examined to provide valid breed comparisons. Although the preference for the Zebu breed was low in this area (7.59%), their positive attributes of easy-keeping cattle with high resistance to disease, better adaptation to harsh climates and powerful draught abilities were still anecdotally recognized by those owners who had them. Breed was negatively associated with natural log of milk production as an ordinal variable in the univariable regression analyses, according to the order of breeds presented above (p<0.05).

For this study, the largest proportion of cows (45.6%) was relatively young between two and five years of age. Age was not recorded as one of the major reasons farmers culled their milking cows, with the oldest cow encountered in the area being 17 years old. There were 37.3% of cows between the ages of 5 and 8 years, while 13.3% of the cows were over eight years old. These age trends were seen as a result of heifers taking a long time before they reached breeding sizes due to nutritional deficiencies that slowed their growth rates (Makau et al 2018 n.d.). Younger animals that were less than five years old produced an average of 6.48 kg/day of milk and this increased to 7.05 kg/day for the middle-aged cows (5-8 years) and then the mean dropped (6.15 kg/day) for cows older than 8 years old. This pattern follows the natural trend of milk production when cows are expected to reach maximum production around 5-6 years of age and at the third parity (Lee and Kim 2006).

The lactating cows had an overall mean and median DIM of 300 and 243 days, respectively, and were categorized into various lactational stages (e.g. early, mid, late, extended, and very long, for cows with DIM of < 100, 101-200, 201-300, 301-400, and >400 days). There were 44, 78, 33, 53 and 108 cows in early (13.9%), mid (24.7%), late (10.4%), extended (16.8%), and very long lactation (34.2%), respectively. DIM was negatively associated with natural log of milk production as a continuous variable in the univariable regression analyses (Table 1).

The 314 lactating cows that were palpated were categorized into various reproductive states (e.g. anestrus, cycling, possible early pregnancy (no diagnosis on rectal examination), and pregnant confirmed by rectal examination). At the time of the study, 43.4% of the lactating cows were bred and/or pregnant, with 28.7% of the cows being confirmed to be over three months pregnant. According to ovary palpation findings, 30.6% of the lactating cows were in an anestrous phase at the day of the rectal examination, with no palpable structures identified from both ovaries. This group of cows had a mean of 201.2 DIM. The cows that were cycling and non-pregnant (n=74) had a mean of 304 DIM. Since these cows were not yet pregnant, these estimates of days open were expected to increase until the cows conceived. With the poor records kept by farmers, it was not possible to determine calving intervals or days open for previous lactations. It was also not possible to reliably determine first service conception percentages or number of breeding per conception. Cow reproductive status was positively associated with natural log of milk production as an ordinal variable in the univariable regression analyses (Table 1), according to the order of states presented above.

The proportion of all the milking cows currently with subclinical mastitis based on CMT > 1 was 44.0%. Cow current mastitis status was negatively associated with natural log of milk production as a dichotomous variable in the univariable regression analyses (Table 1).

The overall mean and median weight of all the milking cows was 388.4 and 362.0 kg, respectively. There were 5.38%, 64.6%, 28.1% and 1.89% cows that weighed <250, 251-400, 401-550, and over 550 kg, respectively. Cow weight was positively associated with natural log of milk production as an ordinal variable in the univariable regression analyses (Table 1).

The overall mean body condition score (BCS) for the lactating cow was 2.44 ± 0.31. No cow was recorded to have a score of 1 or over 4. A majority of the cows (59.4%) had a body condition score less than or equal to 2.0, which is below the desired body condition score, and this could have been due to the time the cross-sectional study was carried out, with many cows being examined months after the most recent dry season, likely leading to insufficient quantities of low quality feed being available to most farmers for feeding. Fisher’s exact test revealed that there were strong differences in body condition for different reproductive status groups, and the body condition differed in different lactation stages. There were also differences (p=0.004) in BCS in cows on those farms that had received training on dairy husbandry (BCS=2.49) versus those who had not received training (BCS=2.26). Imparting knowledge on the farmers was done through farmer training days by extension officers. When the farmers were asked about the topics on which they had received training, cow feeding regimes seemed to be the most common topic (23.1%) that most farmers could recall.

Factors associated with milk yield

In the first step of the modeling of factors associated with the natural log of milk yield, nine variables were found to be associated with the outcome variable at (p<0.15) when individually investigated (Table 1). The correlation matrix did not indicate any serious correlation among these variables; all correlation coefficients were lower than 0.17.

Table 2 shows the results of the final mixed model: one farm characteristic, four cow variables and one farm management factor were strongly associated with the natural log of daily milk, while controlling for the other variables in the model. Many of the expected factors of milk production were in the final model, and we start the model description with them.

Breed type of the cow was associated with milk yield. The indigenous crosses (e.g. Zebu) showed a 23.7% lower milk yield when compared to exotic crosses, which was the baseline as shown in Table 2. Milk yield within these two breed groups did not differ in the final model when the model controlled for other confounding variables, such as weight, DIM and reproductive status, indicating that these other variables were primarily responsible for breed differences in milk production within the two model categories. These results corroborate other findings in Kenya that the low performance of dairy herds on smallholder dairy farms in the region are associated with the type of breeds kept (Omondi and Njehia 2014). The predominance of exotic crossbreeds in this study is an indication of attempts by these farmers at higher milk production, even though other factors hindered production. Farmers that preferred keeping Jersey cows indicated their preference to a smaller cow that was not feeding as much as the other exotic breeds, even with their perceived low milk production.

Heavier cows were found to have yielded more milk on the day prior to the visit. Cows that weighed over 550 kilograms yielded over twice much milk as those that weighed below 250 kilograms. Heavier cows were more likely to have an adequate body condition (BCS>2.75) and were reflective of good feeding management in terms of quality and quantity of feeds, explaining a higher milk yield. Emaciated cows that weighed less than 250 kilograms had the lowest daily milk yield (3.24 kg/cow/day), that was way lower than the means of all the other weight groups. Body weight changes are also affected by the parity of cows, and higher-parity dairy cows often lose more body weight in early lactation compared to lower-parity cows (Roche et al 2007a). The relationship between parity and post-partum body weight changes could not be explored in this cross-sectional study because parity records were not kept on the cows, but cows estimated to be over eight years of age did weigh more than cows less than five years of age, although there was no difference in milk yield between the three different age groups (p=0.269).

Cows in early lactation had 26.4% more milk than cows in mid-lactation (101-200) days. The differences in milk production increased as we moved farther towards cows in later DIM categories; with 55.1% lower milk production in the cows above 400 days in milk, compared to early lactation cows. When all the categories above 100 DIM were compared to the baseline, all the means were clearly different from the baseline. With non-overlapping 95% confidence intervals between 101-300 DIM and >400 DIM (Table 2), cows in this very long DIM category had lower milk production than the mid-lactation and late-lactation cows. Farmers should rebreed their cows sooner to avoid long DIM to ensure good utilization of the animal’s productive life and better milk production. Long DIM are indicative of animals not coming into noticeable heat, getting served in a timely manner, conceiving and/or retaining a pregnancy. Lactation stage was also associated with milk production elsewhere, as reported by Baul et al (2012).

For this study, reproductive status was categorized into groups, namely anestrous, cycling, early pregnancy and pregnant. Nearly a third (30.6%) of the cows in the present study were anestrus, defined as milking, not pregnant and not cycling at the time of examination. Our study showed an increase in milk produced by the cows in different reproductive status groups compared to those cows that were in early pregnancy, which was set as the baseline. Being a pregnant cow was associated with a modest increase in milk yield when compared to those cows that were in early pregnancy. However, cows that were open and cycling had a 28.9% higher milk yield (p<0.05) compared to cows in early pregnancy. It is hypothesized that open and cycling cows were more likely to be in a positive energy balance, while pregnancy above 3 months of gestation can also draw on energy and protein intake. Unfortunately, during the time of this study, cows were found to have extended days open sometimes over 600 days, a situation that was likely brought about by long periods of drought and poor feed storage and management.

There were a few other variables associated with milk production that were interesting. Feeding dairy meal during the last month of gestation lead to increased milk yield such that 35.5% more milk was obtained from the cows that had been received some extra supplementation with high protein concentrate (dairy meal) during the transition period compared to those that had not received any. It has been demonstrated that supplementing dairy cows with 0.5 to 2 kg of dairy meal concentrate per day before parturition, with increasing amounts as parturition approaches has been associated with cows attaining higher levels of milk production during the early days of lactation (Richards et al 2015; Richards et al 2016). Our results confirm that the impact of this management factor may have a lasting effect beyond the first 2 months of lactation. Farmers should be encouraged to practice this management recommendation.

The percentage of land allocated to dairy feed was positively associated with the cow’s milk yield per day. A 25% increase in the land allocated to growing dairy feeds was associated with a 15.0% increase in milk produced, holding all the other factors constant. Due to a constantly increasing population in this area of Kenya, land holdings per owner have decreased by more than half over the past few decades, mainly because of subdivision through family inheritance (Bebe et al 2003). Farmers indicated owning a mean of 2.04 acres of land, leading to competition between growing food for people and feed for the cows. The study showed that with more land allocated to growing feed for their cows, more milk yield could be obtained.

Another factor that has been found to affect the amount of milk produced by cows in the tropics is suckling calves. Some farmers in the tropics still practice restricted suckling in any of its three forms namely: (i) the calf may initiate milk letdown, the cow is milked and the calf sucks residual milk; (ii) the calf is allowed to suck one quarter; or (iii) the calf may suck the residual milk once milking is completed. Restricted suckling has been associated with many advantages over bucket rearing, including increased milk production, increased persistence of lactation and extended lactation, reduced incidence of mastitis, and increased calf growth and survival (Preston and Vaccaro 1989; Little et al 1991; Agyemang et al 1993; Msanga and Bryant 2004; Juhlin 2013). The greatest disadvantage of this practice is said to be its adverse effects on reproduction (Little et al 1991). In the current study, the aspect of restricted suckling was not explored as farmers rarely allowed calves to suckle cows other than for colostrum. Calf rearing and calf management in this study population was described and reported in a separate publication (Makau et al 2018 n.d.).

The intraclass correlation of 0.246 indicated that there was substantial correlation of cows within farms, confirming the need to adjust for clustering of cows within farms using a random herd effect. There was no confounding of model variables among the other variables not in the final model, and no interaction between model variables. The R² of the final model was 0.468, suggesting that 47% of the milk variation was explained by the model.

The quartile plot of the standardized residuals did not indicate any serious deviations from the normal distribution, and the residual plot did not reveal any serious concerns after the data were log transformed. Based on the residual and leverage diagnostics, farm 43 had somewhat high values, though within the acceptable range of 3 and -3. The magnitude and the influence of the residuals for farm 43 did not reveal any problems; when the model was analysed without this farm, there was little change in the variables, and thus farm 43 was retained in the final model.

Research limitations included a language barrier, especially with aged farmers who could only communicate in the native Kimeru language. This needed an interpreter who was fluent in the native language to relay the message and convey the respondents’ answers to the research team. Suspicion and mistrust was also noted among some respondents, particularly with details surrounding their personal life, and this got in the way of data collection. It was however mitigated by assuring them that the information given would be treated with utmost confidentiality, respect and professionalism. There were a few uncooperative and unfriendly respondents, but this situation improved when word went around the community about the project and its objectives.

Data quality was considered to be good by the researchers because collection was carried out by a well-trained team, and the questionnaire used was adopted from a previous study carried out in a different part of Kenya, and thus it had been pretested and modified. Participants of this study were randomly selected to avoid any selection bias, and of the 200 farmers that were selected for the study, they all agreed to participate and provided the requested data. Physical, clinical and rectal examinations were done by qualified veterinarians and veterinary students under the supervision of veterinarians.

Since this was a cross-sectional study, we cannot use results herein to determine causality of the model factors, but the results obtained were used to as a guide in the randomized control trials that were to follow the project. A more detailed cohort study or trial is recommended to test hypothesized model factors, to document and examine all the changing cow and management factors over time, and to provide the necessary evidence for recommendations for farmers, in turn improving the output from their dairy enterprises.

Conclusions

• The mean and median milk yield for the study cows in this study was 6.7 ± 0.23 and 6.0 kg/cow/day respectively.
  
  
• The principal farmer was female in 52.5% of the farms, and had either none or only the basic level of formal education (grade 8), leaving a need of training in area that would lead to improvement of the dairying enterprise.
  
  
• A third of the farmers practiced crop farming as a way to supplement their income. As the mean total land holdings owned by the respondents was small (2.04 acres), the farmers were in dire need to find ways to support their families, including improved efficiency with the land they had.
  
  
• Reproductive performance among the study cows was sub-optimal, with over half of the milking cows being open at the time of the study and had an average of 253 DIM among these open cows. Only 28.7% of the milking cows were confirmed pregnant at the time of the study.
  
  
• The sub-optimal milk production of the dairy cows recorded in this area was associated with cow-level factors including; breed type, weight, reproductive status, and lactation stage as well as farm-level factors including; education levels of the male farmer, percentage of land allocated to growing fodder for dairy cows, and whether or not the farmers provided high protein supplements to dry cows before parturition.
  
  
• Extension services for training SHD farmers on best management practices associated with record-keeping, dairy cattle management (especially nutrition and reproduction), and fodder growing and conservation should be improved to enhance milk production on these farms.

Acknowledgements

We are grateful to the primary funding program for this research, the Canadian Queen Elizabeth II Diamond Jubilee Scholarships (QES) which are managed through a unique partnership of Universities Canada, the Rideau Hall Foundation (RHF), Community Foundations of Canada (CFC) and Canadian universities. This research is made possible with financial support from the Government of Canada, provincial government and the private sector. We also acknowledge the large contribution made by volunteers and staff of Farmers Helping Farmers, a non-governmental organization - their existing relationships and agricultural efforts and inputs provided a strong foundation for the work and the entry point to the Naari community. As well, the support of the Naari Dairy Cooperative Society and the cooperation of the dairy farmers made it all possible.

References

Agyemang K, Clifford D and Little D A 1993 An assessment of the biological and economic efficiency in conversion of milk to growth in N’Dama calves. Animal Production, 56(02), 165–170.

APSK 2015 Proceedings of the APSK Annual Scientific Symposium, (April).

Asoka N, Thuo G W and Bunyasi M M 2013 Effects of Population Growth on Urban Infrastructure and Services: A Case of Eastleigh Neighborhood Nairobi, Kenya. Journal of Anthropology & Archaeology, 1(1), 41–56.

Baul S, Cziszter L T, Acatincăi S, Gavojdian D, Tripon I, Erina S and Răducan G G 2012 Phenotypic Correlations among Milk Yield and Chemical Composition per Normal Lactation in Romanian Black and White Breed, 45(2), 275–277.

Bebe B, Udo H M, Rowlands G and Thorpe W 2003 Smallholder dairy systems in the Kenya highlands: cattle population dynamics under increasing intensification. Livestock Production Science, 82(2–3), 211–221.

Dohoo I R, Martin S W and Stryhn H 2009 Veterinary Epidemiologic Research. VER Incorporated.

Government of Kenya (n.d.) Meru North District Development plan. From meru.go.ke/lib.php?com=114&&res_id=618

Hare E, Norman H D and Wright J R 2006 Trends in Calving Ages and Calving Intervals for Dairy Cattle Breeds in the United States. Journal of Dairy Science, 89, 365–370. http://doi.org/10.3168/jds.S0022-0302(06)72102-6

IFAD 2006 small holder dairy commercialization programme. Retrieved from http://www.sdcp.or.ke/index.php

Juhlin L 2013 Traditional milk production in cattle in a semi-arid area in Kenya. Retrieved from http://www.slu.se/husdjurmiljohalsa

Kristula M A, Curtis C R, Galligan D Tand Bartholomew R C 1992 Use of a repeated measures logistic regression model to predict chronic mastitis in dairy cows. Preventive Veterinary Medicine, 14(1 2), 57 68; 18 ref.

Lee J-Yand Kim I-H 2006 Advancing parity is associated with high milk production at the cost of body condition and increased periparturient disorders in dairy herds. Journal of Veterinary Science, 7(2), 161–6.From http://doi.org/10.4142/JVS.2006.7.2.161

Little D A, Anderson F M and Durkin J W 1991 Influence of partial suckling of crossbred dairy cows on milk offtake and calf growth in the Ethiopian highlands. Tropical Animal Health and Production, 23(2), 108–114. http://doi.org/10.1007/BF02361194

Makau D, VanLeeuwen J A, Gitau G K, Muraya J, McKenna S L, Walton C and Wichtel J J (n.d.) Animal and management factors associated with weights gain in dairy calves/heifers on Smallholder Dairy farms in Kenya. Preventive Veterinary Medicine (accepted for publication).

Melaku A and Gurmessa J 2012 Effect of Lactation Stage, Pregnancy, Parity and Age on Yield and Major Components of Raw Milk in Bred Cross Holstein Friesian Cows. World Journal of Dairy & Food Sciences, 7(2), 146–149.

Moran J 2005 Tropical dairy farming : feeding management for small holder dairy farmers in the humid tropics. Csiro publishing. From http://www.publish.csiro.au/book/5126/

Msanga Y N and Bryant M J 2004 Productivity of crossbred dairy cows suckling their calves for 12 or 24 weeks post calving. Tropical Animal Health and Production, 36(8), 763–73.

Mutarari M P 2010 Influence of socio-economic factors on rural poverty: A case of Uringu division of Meru-North district in Eastern province, Kenya. University of Nairobi.

Nicholson M and Butterworth M H 1986 A guide to condition scoring of zebu cattle. Retrieved from https://cgspace.cgiar.org/handle/10568/49674

Njarui D M G, Gatheru M, Wambua J M, Nguluu S N, Mwangi D M and Gichene A K 2011 Development and up-scaling of green manure legume technologies in Kenya. Kenya Agricultural Research Institute. From http://www.lrrd.org/lrrd23/5/njar23111.htm

Norman H D, Hutchison J L and Miller R H 2010 Use of sexed semen and its effect on conception rate, calf sex, dystocia, and stillbirth of Holsteins in the United States. Journal of Dairy Science, 93(8), 3880–3890.

Odero-Waitituh J A 2017 Smallholder dairy production in Kenya; a review. Livestock Research for Rural Development , 29(7).

Odima P A, McDermott J J and Mutiga E 1994 Reproductive performance of dairy cows on smallholder dairy farms in Kiambu District, Kenya: Design, methodology, and development considerations., 18(2), 366–368.

Omondi, W and Njehia B 2014 Herd Characteristics on Smallholder Dairy Farms in Western Kenya. Journal of Animal Science Advances, 4(8), 996. From http://doi.org/10.5455/jasa.20140827111904

Owen E, Smith T, Kitalyi A and Jayasuriya N 2005 Livestock and wealth creation: improving the husbandry of animals kept by resource-poor people in developing countries. (1st edition). Nottingham University Press.

Preston T R and Vaccaro L 1989 '' Dual purpose cattle production systems", New Techniques in Cattle Production, (C.J.C. Phillips Ed.) Butterworths and Co. London.

Radostits O M, Leslie K E and Fetrow J 2001 Herd health food animal production medicine (3rd edition). W.B Saunders Company Philadelphia.

Richards S, VanLeeuwen J A, Shepelo G, Gitau G K, Wichtel J, Kamunde C and Uehlinger F 2016 Randomized controlled trial on impacts of dairy meal feeding interventions on early lactation milk production in smallholder dairy farms of Central Kenya. Preventive Veterinary Medicine. From http://doi.org/10.1016/j.prevetmed.2016.01.006

Richards S, VanLeeuwen J A, Shepelo G, Gitau G K, Wichtel J, Kamunde C and Uehlinger F 2015 Associations of farm management practices with annual milk sales on smallholder dairy farms in Kenya. Veterinary World.

Roche J R, Macdonald K A, Burke C R, Lee J M and Berry D P 2007 Associations among body condition score, body weight, and reproductive performance in seasonal-calving dairy cattle. Journal of Dairy Science, 90(1), 376–91.

Romney D, Kaitho R, Wambugu M, Chege L, Omore O, Staal S and Wanjohi P 2000 Technology development and field-testing: Access to credit to allow smallholder dairy farmers in central Kenya reallocate concentrates during lactation. In Conference of the Egyptian Society of Animal Production, Alexandria, Egypt. (pp. 133–139).

Rufino M C, Herrero M, Van Wijk M T, Hemerik L, De Ridder N and Giller K E 2009 Lifetime productivity of dairy cows in smallholder farming systems of the Central highlands of Kenya. Animal, 3(07), 1044–1056. From http://doi.org/10.1017/S1751731109004248

Swai E S, P Kyakaisho and Ole-kawanara M S 2007 Studies on the reproductive performance of crossbred dairy cows raised on smallholder farms in eastern Usambara mountains, Tanzania. Livestock Research for Rural Development, 19(5).

Thornton P K 2010 Livestock production: recent trends, future prospects. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 365(1554), 2853–67. From http://doi.org/10.1098/rstb.2010.0134

VanLeeuwen J A, Mellish T, Walton C, Kaniaru A, Gitau R, Mellish K and Wichtel J 2012 Management, productivity and livelihood effects on Kenyan smallholder dairy farms from interventions addressing animal health and nutrition and milk quality. Tropical Animal Health and Production, 44(2), 231–238.

Wakhungu J W, Mosi R O and Charles A B 2000 Dairy Cattle breeding policy for Kenyan smallholders: evaluation based on demographic stationary state productivity model. In Proc. Dairy Cattle Breeding in East Africa Sustainable Artificial Insemination Service, KARI HQs (pp. 17–18).

Received 27 August 2018; Accepted 6 September 2018; Published 1 October 2018

Go to top