Micropatterns induced by sheep grazing, were studied in three consecutive years in a Festuca rubra-dominated salt marsh in a grazing trial with five different stocking rates (0, 1.5, 3, 4.5 and 10 sheep ha(-1)). The micropatterns were formed by a mosaic of short and tall F. rubra stands on a scale of square decimeters. Permanent transects of 2 m x 10 m were used to study the stability of these patterns, and to analyze interactions between the vegetation, the marsh elevation and the sheep. Micropatterns occurred only in the lightly to moderately grazed paddocks (1.5-4.5 sheep ha(-1)) with the highest spatial diversity in the 3 sheep ha(-1) transect. When grazing was excluded, micropatterns did not develop; nor did they develop in the traditionally and most intensively grazed paddock (10 sheep ha(-1)). Detailed observations in one year showed that crude-protein content did not differ between green leaves from the short and tall stands, whereas in vitro digestibility was slightly higher in the short stands. In the same year, tiller density and length of full-grown leaves increased substantially in both stands from May to September. At the same time, sheep preference shifted from tall to short stands, which suggests an interplay between intake rate and digestibility in the sheep selectivity.
Seven years after establishment of the grazing trial, the 10 sheep ha(-1) transect still showed a smooth relief typical of the starting point of the other transects. These transects developed a more hummocky topography, with the highest spatial diversity occurring on the 1.5 sheep ha(-1) transect. Marsh elevations were on average up to 3 cm lower in the short than in the tall stands, which indicates that the somewhat lower-elevated patches were grazed more intensively than the higher-elevated patches.
In most cases, micropatterns changed from one year to the other, probably due to weather fluctuations. The incidence of tall stands was influenced by the rainfall balance. If the incidences of both the short and the tall stands were around 50%, however, the micropatterns showed a clear correlation with the marsh elevation. The rainfall balance seemed therefore a decisive factor for a possible correspondence between micropatterns in two consecutive years. Elevation differences were so subtle that greater than average sedimentation during a winter season could change the elevation pattern. Hence both rainfall balance and winter sedimentation counteracted the stability of the micropatterns. During our three-year study period, micropatterns were only stable in one out of six possible paired comparisons. This low micropattern stability contrasts with other studies in inland environments, which shows that in more dynamic environments, abiotic processes are likely to overrule summer grazing in determining vegetation patterns.