Trials, reflecting the feed needs in dry Mediterranean environments of small-ruminant production systems based on barley, were established at two sites in Syria in 1982. They compared various 2-course rotations of barley with feed legumes, fallow or more barley. This paper summarizes a 14-year sequence of results from an incomplete factorial combination of four rotations (B-V, B-L, B-F, B-B) of barley with vetch (Vicia sativa), lathyrus (Lathyrus sativus), fallow, and barley, with two fertilizer regimes, zero control and biennial NP applied to the barley phase, in terms of long-term mean yields, production stability and yield trends over time.

On a 2-year rotational basis, most barley was produced by barley-only rotations, and differences between B-F and B-B were small; but, in terms of total biomass production, feed legume rotations (B-V and B-L) outyielded barley-only rotations by 29% at one site and 19% at the other. Responses to biennial fertilization were large but did not interact significantly with rotation treatment. The crude protein status (%N) of barley grain and straw was strongly determined by seasonal rainfall, but that of the grain could be enhanced, irrespective of rainfall, by a preceding feed-legume crop; and, altogether, the total mean crop nitrogen output of legume-based rotations exceeded that of barley-only rotations by 80% and 64% at the two sites. The inclusion of legumes thus enhances both quantity and quality of feed production.

Annual yield fluctuations, attributable mainly to rainfall difference, were greater at the drier site. No consistent effect from fertilizer was observed, but at the wetter site rotation differences were appreciable, with B-F rotation giving the most stable yields. A number of time trends in yield values were tentatively identified. On a relative basis, some widening over time of the gap between fertilized and unfertilized treatments was observed in feed-legume yields at both sites and barley yields at the wetter site; over 14 years, yields in unfertilized plots had apparently declined relative to those receiving biennial NP. But, apart from a probable decline in lathyrus productivity compared to that of common vetch, changes in relative yield performance between rotations were difficult to detect. Regression models developed to describe absolute yield trends indicated a real decline over time in barley grain yields in continuous barley (B-B) at both sites and in unfertilized plots of all four rotations at the wetter site.