||1. The species composition and spatial distribution of small insects (Hemiptera, Coleoptera, Lepidoptera) and arachnids (Araneae, Opiliones, and Pseudoscorpiones) were investigated in three indigenous, upland grasslands identified as the National Vegetation Classification Festuca–Agrostis–Galium typical subcommunity (code U4a), Festuca–Agrostis–Galium, Vaccinium–Deschampsia subcommunity (code U4e), and Nardus stricta species-poor sub-community (code U5a), on which grazing management was manipulated experimentally.
2. Two hypotheses were tested that predicted arthropod diversity in upland grasslands. The habitat heterogeneity hypothesis predicts that the species number and abundance of arthropods will have an asymptotic relationship with increasing numbers of plant species and greater structural heterogeneity in the vegetation. The symbiosis between patches hypothesis states that the species number and abundance of arthropods will express a unimodal relationship with the grain size of sward patches created by grazing. The sward patches must be large enough to be apparent to, and support populations of, arthropods, but small enough that interspersed tussocks provide shelter from weather and a deterrent to disturbance by grazers.
3. The hypotheses were tested by sampling arthropods from the geometrical patterns represented by the individual tussocks and intermediate sward components of three indigenous grasslands produced by different grazing treatments. Paired samples of arthropods were taken by motorized suction sampler, the first of the pair from the grazed sward and the second, the accumulated samples from the surrounding triad of tussocks (U4a and U5a grasslands) or hummocks (U4e grassland). The paired samples were taken from six randomly-selected locations across both replicates of each of the grazing treatments.
4. Arthropod species composition and abundance were compared between the paired sward and tussock samples and in turn with measures of the vertical and horizontal components of vegetation structure, i.e. the variance in vegetation height per unit area and the area covered by tussock compared with sward.
5. There were consistently more species and a greater abundance of arthropods associated with tussocks than with swards and the average species number and abundance for the combined pair of samples declined with increased grazing pressure. The relationship between vertical and horizontal components of vegetation structure and the species number and abundance of selected arthropods was asymptotic as opposed to unimodal, supporting the habitat heterogeneity hypothesis, rather than the symbiosis between patches hypothesis.
6. Small and relatively sedentary insects and arachnids are more sensitive to grazing intensity and species of grazer in these upland, indigenous grasslands than are larger Coleoptera and Araneae, which respond less directly to varied grazing management. The overall linear reduction of small herbivorous and predatory arthropods with increased grazing intensity was buffered in grasslands with substantial tussock patches.