Understanding and representing the transformation of agricultural production systems has attracted increasing interest due to its importance for identifying drivers of changes and coping mechanisms in response to global challenges. These challenges are all the most pressing in North African countries exposed to a dramatic reduction in rainfall and increasing temperatures that affect sustainability in more than half of this semi-arid territory. This paper describes an improved way to understand such transformations through a cross-cutting analysis of crop–livestock system indicators over a period of 10 years in one community in Southern Tunisia. Our methodology is divided into four steps: (i) systems identification, (ii) indicator-based assessment of system crop–livestock sustainability, (iii) hierarchical clustering to identify sustainable intensification (SI)-based farm types and, finally, (iv) analysis of trajectories of these farm types. Results showed that the sustainability of the systems studied increasingly depends on diversification rather than intensification, which dominated in the 2000s. However, diversification has not necessarily improved socio-economic viability. Over the 10-year period, results revealed a dramatic increase of almost 50% in the population of small-scale farms whose viability depends on a range of on- and off-farm activities to meet the short-term needs that correspond to a buffer strategy. Additionally, the dominant SI processes were shown to be mostly based on diversification to livestock activities with both milking and fattening. Our holistic and timeline approach to system transformation makes it possible to account for sustainability between (systems) generations, which will be highly needed in future discussions about sustainability.

The Maasai of Kenya are typical of many pastoral communities in that they rely on milk as a major part of their diet. Milk production in three herds of Maasai cattle was studied from 1991 to 1996. Weekly measurements of milk extracted for human consumption were used to estimate lactation length, total offtake and mean daily production. Calf live-weight data were used to estimate total milk yield. Least-mean squares regression and analysis of variance tests were adopted to assess the effect of a number of variables on milk production parameters.

Data from a total of 650 lactations and 383 calvings were collected. Mean daily offtake was 891 ml. The length of lactation of frequently milked cows was 372 (s.e. 7·96) days. Total lactation offtake was estimated at 305 I. Herd, parity level and year of calving differences were observed. Multiple regression analyses indicated that some of the variation in daily offtake could be explained by stage of lactation, season of milking, herd, parity and year of calving. An average total yield of 2·73 I/day was estimated for the first 90 days of lactation. Of this, approximately 0·4 was used for human consumption. Overall, seasonality was seen as the major variable influencing production.

The size and composition of three herds belonging to Maasai pastoralists were monitored for more than 5 years (1990 to 1996). Animals were categorized as either suckling calves, weaned heifers, weaned males or breeding females. The dates of entries and exits were used to estimate the total number of observed animal days for each category. Fertility rates of breeding females and mortality and disposal rates for each category were estimated using animal days as the denominator. Herd differences were tested using contingency tables. Age to first calving and calving intervals were estimated and examined using analysis of variance tests.

The patterns of births and deaths were seasonally influenced. The mean annual calving rate of all breeding females was 65·6% and for females excluding first calvers was 46·9%. No herd effects were observed. The mean observed interval between calvings was 609 days and the calving interval, calculated from the parturition rate, was 649 days with no herd differences. Mean age at first calving was approximately 4 years with no herd differences. Mean annual mortality rates were 8·9% for breeding cows, 7·8% for weaned males, 6·7% for weaned heifers and 22·1% for calves. The mean annual culling rate of cows was 10·8% and the sale rate of heifers and weaned males was 12·9% and 41·5% respectively. The mean age at disposal of heifers was 655 days and of steers was 801 days. Herd sizes were shown to fluctuate annually with no obvious trend between herds. The main determinants of production identified were the effect of seasonally poor nutrition on the rate of reproduction and the effect of season, herd mobility and disease on mortality.

The study describes production levels that can be used to predict future changes to the system. The results are discussed in terms of the factors influencing and their relevance to pastoral development. The findings are compared with those observed in other, similar areas.