Farm mechanization promises to help raise labor productivity and reduce the heavy toil of farming on the world’s millions of smallholder farms, hence contributing to socioeconomic development in the Global South, in particular in Africa. While mechanization is therefore high on the African development agenda, there are heavy – at times dogmatic – debates on which technological pathway toward farm mechanization – animal traction, two-wheel tractors, and four-wheel tractors – should be supported by African governments and development partners. One discussion area relates to the future of animal traction. Proponents see a continued scope for the use of draught animals, whereas opponents see animal traction as old-fashioned and see a potential to leapfrog this mechanization stage. There are also debates on the potential of two-wheel tractors, with proponents arguing that such walk-behind tractors are more affordable and suitable for smallholder farmers, and opponents believing that such tractors lack efficiency and power and still come with a high drudgery. This paper argues that there are no blueprint answers on which technological pathway is ‘best’ but only answers on which one ‘best fits’ the respective conditions. Based on this premise, this paper introduces a ‘best-fit’ framework that allows for assessing the comparative advantages and disadvantages of the three technological pathways in different agroecological and socioeconomic conditions. The results suggest that all three forms of mechanization are associated with areas where they ‘best fit’. All three farm mechanization pathways hinge on public policies and investments to create an enabling environment for private markets, as, ultimately, innovation processes should be market driven. The ‘best-fit’ framework enables governments and development partners to focus efforts to support farm mechanization on solutions that ‘best fit’ their country’s farming systems and not on those that are politically most attractive, thereby contributing to sustainable agricultural mechanization and development.

This study examined whether work applied at different stages of lactation had any effect on reproduction of cows under two feeding regimes. Twenty-four F1 crossbred dairy cows (12 Friesian × Boran and 12 Simmental × Boran) were allocated one of two diets (H + 3: natural pasture hay + 3 kg concentrate per day; and H + 5: natural pasture hay + 5 kg concentrate per day) and one of three work treatments — starting 45 days (D45), 90 days (D90) and 135 days (D135) post partum — using a principal component analysis score based on milk yield, live weight, calving interval and parity. Cows pulled sledges for 50 days (pull = 108 N per 100 kg live weight). Over the experimental period of 315 days, cows on diet H + 5 travelled a similar distance and produced amounts of fat-corrected milk (FCM) similar to those of cows on diet H + 3. Total intake of dry matter per kg live weight 0·75 was higher for cows on diet H + 5 than for cows on diet H + 3 and was similar across work times. Hay dry-matter intake was greater for the H + 3 group than for the H + 5 group at 180 days post partum and thereafter. Cows on diet H + 5 lost less weight in early lactation and gained more in mid and late lactation than cows on diet H + 3. During the work period, live-weight change was similar across diets but it was different between work treatments D45 and D135. The interval from calving to conception decreased by 63 and 101 days when start of work was delayed from D45 to D90 and from D45 to D 135, respectively. Output/input ratios of metabolizable energy equivalents were 0·35 for H + 3 and 0·37 for H + 5 diet, and 0·34, 0·37 and 0·40 for work times D45, D90 and D135, respectively. These results indicate that work started in early lactation significantly increased days to conception and decreased overall productivity of lactating working cows. Farmers must weigh the relative importance and cost of delayed ploughing against those of delayed oestrus or against the cost of borrowing draught power.

The work performance of two teams of four donkeys (heavy, 680 kg and light, 460 kg) and one pair of Jersey crossbred oxen (646 kg) was compared when they ploughed 4 hi day on four types of soil (clay, redsoil, sandy soil and sandy clay) using two types of plough, a conventional ox plough (40 kg) and a lighter prototype, the ‘Walco’ plough (32 kg) on an experimental farm. Work parameters were also measured with farmers’ cattle and donkey teams ploughing on f arms in Matobo and Nkayi districts. Working speed, power and effective field capacity (ETC) were higher for the ox-team (1·03 m/s, 920 W and 14·5 h/ha for the conventional plough and 0·99 m/s, 745 W and 13·9 h/hafor the Walco plough) and the heavier donkey team (0·87 m/s, 689 W and 14·2 h/hafor the conventional plough and 0·87 m/s, 787 W and 17·3 h/hafor the Walco plough) than for the lighter donkey team (0·59 m/s, 461 W and 22·1 h/hafor the conventional plough and 0·64 m/s, 445 W and 23·4 h/hafor the Walco plough). Expressed as a proportion of live weight or metabolic live weight there were no significant differences in draught forces exerted between teams but power output per unit live weight was greater in the ox-team than in the light donkey team but similar to that in the heavy donkey team. The Walco plough required a lower force (742 N) to operate than the conventional plough (816 N) but apart from this did not have any marked advantages over the conventional plough. On-farm, team sizes of donkeys varied from three to seven animals (team weight 340 kg to 1007 kg) and cattle team sizes from two to four animals (team weights 558 to 1709 kg). Regardless of team number, the heavier teams tended to out-perform the lighter teams (speed range 0·63 to 1·08 m/s, power 395 to 1136 W, EFC 9·1 to 25 h/ha)) with one exception, a well trained team of two oxen (team weight 879 kg, speed 1·02 m/s, power 775 W, EFC 9·1 h/ha). Donkeys tended to plough at a slower pace than oxen, with a lower power output, although when weight differences between teams were equalized (four heavy donkeys compared with two oxen), then there was little to chose between the species. Results suggested that teams of three or more donkeys can effectively be used for ploughing on the soils tested. The results highlighted the importance that team live weight and training/experience have in determining work performance.

A feeding system, draught cows, disease/work interactions and animal power introduction in a farming system, four areas of recent research on draught ruminants which are important to future development of animal power, are reviewed. A new feeding system for draught animals is described which enables food requirements and the effects of work on live weight and milk production to be calculated.

Recent data on the energy cost of walking are appraised. Research on working cows, mainly in Ethiopia, has shown that undernutrition has a greater effect on milk yield than work, which has a transient effect. The length of the post-partum anoestrous period increases with decrease in body condition. Body-weight loss increases with increasing work load. It is suggested that dairy cows delay conception by 1 day for every day of work done. Work has little effect on food intake or digestive parameters. Although it is associated with an overall increase in food intake of cows, even of un-supplemented forage diets, the increase is not sufficient to meet all the extra energy needs for work. Food intake of both working and non-working cows increases during lactation.

Disease limits the working capacity of draught animals and work can exacerbate disease. These effects were investigated using Trypanosoma evansi in Indonesia and T. congolense in The Gambia. In both studies, infected animals were able to do much less work than non-infected ones and the severity of the effect depended greatly on the strain of trypanosome used. In general, increasing the plane of nutrition did not ameliorate the effects of the disease, nor in the Gambian study did it prevent loss of appetite in infected animals.

The technical and agronomic innovations necessary for the introduction of animal power into an inland valley region of central Nigeria are described and some of the sociological implications discussed.

The heart rate (fH) and the energy expenditure (EE) of seven Hintenvaelder (Bos taurus) draught oxen and three zebu (Bos indicusj oxen were measured, while the animals were standing, walking and pulling different loads. Linear regressions for all animals relating EE to fH were highly significant (P < 0·001). The standard errors of the estimate expressed as a percentage of the mean EE (PE) ranged from ±6·7% to ±10·5%. Two animals with PE ±13·7% and ±17·1% were beyond that range. One year later, fH and EE were measured on six of the original seven Hinterwaelder oxen while the animals were standing and walking on a treadmill, on the level and at gradients of 3%, 6% and 9%. In the two experiments mean slope and mean intercept of the regressions of EE on fH were not different (P > 0·05). Irrespective of the kind of work (draught work or lifting work),fH allows a reliable prediction to be made of the EE of working oxen.

Two experiments were conducted to investigate the relationships between work and intake and digestion of food by draught oxen given millet stover. In the first experiment, intake of millet stover, water intake, live weight, plasma concentrations of triiodothyronine, thyroxine and urea-nitrogen were measured in 18 animals that worked 0, 2 or 4 h/day in sequence during three 3-week experimental periods. Digestibility and rate of passage of food residues through the digestive tract were measured in a second experiment on 12 animals working either 0, 2-5 or 5 h/day in sequence during three 2-week experimental periods. Feeding behaviour was monitored on six animals working either 0, 2-5 or 5 h/day. Work did not affect intake of millet stover, apparent digestibilities and the rate of passage of digesta through the gastro-intestinal tract. This suggests that the nutrient supply from intake of roughages by ivorking oxen is unlikely to be sufficient to compensate for the extra energy expended during work. Food intake was affected by the quality of the millet stover offered. The level of intake of millet stover ivas proportional to the amount of leaves in the stover. Food intake increased also as work progressed. However, animals mobilized their body reserves to perform work. Animals consumed more water on working days than on days they were at-rest in shade. The heat stress that working animals were subjected to did not appear to interfere with their digestive function.

The Oxylog, a portable breath-by-breath gas analyser, was used on seven animals to determine standing metabolic rate, energy cost of walking on soils of different consistencies and efficiency of work ploughing and carting. The average standing metabolic rate of animals was 5·63 (s.e. 0·12) W/kg M00·75. The consistency of the soil on which animals worked had a marked effect on their energy cost of walking which was 1·59 (s.e. 0·069) on unploughed soil, 2·15 (s.e. 0·084) on ploughed soil and 1·0 (s.e. 0·10) J/m per kg live weight on laterite tracks. The efficiency of ploughing sandy soils (i.e. ratio of work done to energy used for work) was 0·32 and was not significantly different from the efficiency of carting with different loads. The efficiency of doing work was not influenced by the type of work performed, the draught force exerted or the walking speed.

Eighteen oxen were allotted to three treatment groups according to their body condition: poor, medium and good. Work output, speed, live weight and body condition were measured during 7 weeks when animals worked 4 days/ week, 4 h/day, pulling loads equivalent to 12-5 kgf/100 kg live weight. The animals were given millet stover ad libitum during hours they did not work plus 10 g/kg M of a concentrate mix. Work did not influence intake of millet stover. However, food intake improved as work progressed and animals in bad condition ate more millet stover than animals in good body condition. Work performance was affected by live weight but not body condition. Live-weight losses did not have a detrimental effect on work performance. Power output improved during the course of the experiment while animals were losing weight Animals in all treatment groups lost body weight during the 7 weeks of work but weight losses were more pronounced in oxen in good than in poor body condition. At the end of the working period, animals were put on natural pastures without supplementation. It took 4 weeks for animals in poor and medium body condition and 6 weeks for animals in good body condition to reach their pre-work live weight.

Heart rate, ventilation measurements and gas exchange were studied in seven Hinterwaelder (Bos taurus) (494 (s.e. 16) kg) and five zebu (Bos indicusj oxen (516 (s.e. 60) kg), while the animals were standing, walking and pulling different loads. During standing, the heart rate, respiration rate, ventilation volume per min, total carbon dioxide production, total oxygen consumption and oxygen consumption per heart beat were lower in the zebu group than in the Hinterwaelder group (P < 0·05). The higher values for the Hinterwaelder were probably due to their higher metabolic rate, because of their younger age, higher level of feeding and better quality diet. The adaptation of gas exchange to higher work loads was achieved mainly by higher ventilation volumes. The composition of the expired air changed minimally. At similar levels of draught power output, the total energy expenditure minus energy expenditure of walking of the zebu oxen was lower than that of the Hinterwaelder oxen (P < 0·02). On the other hand at similar levels of heat production, the heart rate and ventilation volume per min did not differ significantly between the two breeds (P > 0·05). In spite of the differences in environment, feeding level and draught efficiency, the physiological adaptations made by the two breeds in response to work were similar.

Eighteen F1 crossbred dairy cows (Friesian × Boran and Simmental × Boran) were allocated to one of three diet groups (H: natural pasture hay; H+3: natural pasture hay + 3 kg concentrate; and H+5: natural pasture hay + 5 kg concentrate) using a stratified random sampling procedure, with parity, milk production genotype, body weight and body condition score as blocking variables. Cows on each diet were then allocated to three draught forces (7, 11 and 15 kg draught force per 100 kg live weight) in a cross-over design to investigate relationships between work output, live-weight changes, dry-matter intake (DMI) and milk yield. Each cow worked for 36 days in early lactation (from calving to 90 days) and for a further 36 days in late lactation (from 250 to 340 days post partum). Work output was similar for cows on each of the three diets. Cows on the H diet consumed more hay than cows on H+3 and H−5 diets. Hay and total DMI, milk yield and milk fat were similar across draught forces and during working and resting days when the cows were not pregnant. Similar results were obtained when cows were from 82 to 172 days pregnant. Pregnancy did not affect the ability of cows to perform work at different intensities. During working days cows lost live weight both when pregnant and when non-pregnant. During rest days, non-pregnant cows on diets H, H+3 and H+5 compensated proportionately 0·12, 0·59 and 0·59, respectively, of the live weight lost during working days. Pregnant cows on diet H+3 and H+5 compensated proportionately 0·95 and 1·77 live weight, respectively. Lower total live-weight losses during the pregnant period could be attributed partly to relatively greater DMIs and lower milk production, but also to gestation. Prediction of live-weight change from total DMI above maintenance, milk yield and work output during periods of 6 working days was poor (R2 = 0·18). However, the same parameters explained adequately changes in live weight for supplemented and non-supplemented working cows (R2 = 0·38 and 0·79, respectively) during a period of 90 days.

Forty pregnant F1 crossbred cows (20 Holstein-Friesian × Boran and 20 Simmental × Boran) were assigned to a 2 × 2 (work × diet) factorial experiment as follows: supplemented-non-working (SNW), supplemented-working (SW), non-supplemented-non-working (NSNW) and non-supplemented-working (NSW). Working cows pulled sledges 100 days/year (pull = 350 to 450 N, 4 h/day, 4 days/week). Conception and oestrus at fixed times (200 and 365 days post partum) were analysed using linear logistic models. Proportional hazard models were used for analysing ‘failure’ time data such as time to first oestrus or time to conception. Diet supplementation significantly decreased days to first oestrus and days to conception in non-working and working cows. SW cows had similar reproductive performance to NSNW cows. In supplemented cows, work significantly delayed days to conception. However, by 365 days post partum, conception rate was similar for SNW and SW cows. Body condition at calving significantly affected post-partum reproductive ability of non-working and working cows. Natural grass hay alone could not support potential reproductive ability of crossbred cows. Work output of supplemented cows may be associated with longer calving intervals. The economic trade-offs between longer calving intervals and work output should be examined in detail.

The effect of exercise (walking 10.6 km/day and climbing 480 m/day) on intake of barley straw and body weight changes in 12 pregnant lactating suckler cows was investigated. Three groups of four cows received ad libitum barley straw and 4 kg per cow per day of different supplements. Neither exercise nor supplement had a significant effect on straw intake. Cows gained weight when not exercised, but lost weight during the exercise period. The experiment confirms previous findings that animals performing moderate work and given poor quality roughage, in the short term cannot increase daily intake to meet energy requirements and that intake may decrease when animals work.

Cardio-respiratory and mechanical work output variables were measured for four F1 crossbred dairy cows (average weight 475 kg) over a range of work levels. Resting (standing prior to exercise) heart rates (average 62 beats per min) and respiratory rates (average 21 breaths per min) were similar to those previously reported for cattle of similar weight. Minute ventilation was 0·146 l/min per kg. Resting oxygen uptake (average 0·0057 l/min per kg) was higher than values that have been reported for cattle under more relaxed conditions.

The cows were exercised on an oval track by undertaking two bouts of activity for 500 m with a 5-min intervening recovery period. Walking increased minute ventilation by approximately 60 I/min and heart rate by 30 beats per min. Pulling a loaded sledge increased these variables further to approximately 200 l/min and 135 beats per min at 32 kg/min (530 W), the highest work level studied. The calculated oxygen pulse was 72 ml per heart beat, with maximum oxygen consumption approximately 9 l/min.

Overall ivorking efficiency increased with work level to a maximum of approximately 26% at the highest work level. At this level the cows were pulling proportionately about 0·14 of their body weight.

During periods of recovery after work, the cardio-respiratory parameters varied as would be expected and recovery was usually complete within 7 min.