Oecologia (Berlin) (1986) 70:508-513 Oecologia 9 Springer-Verlag 1986 Community changes following shrub invasion of grassland R.J. Hobbs* and H.A. Mooney Department of Biological Sciences, Stanford University, Stanford, CA 94305, USA Summary. We studied the development of the shrub Bac- charis pilularis ssp consanguinea and its effects on herbs of the annual grassland in Northern California. A series of stands of Baccharis was sampled of ages ranging from 1 yr to >9 yr, representing most of the life cycle of the shrub. In each stand we examined shrub biomass, structure and litterfall. We also determined cover and biomass of all herbaceous species and estimated seed production, seed rain and storage of seed in the soil. Abundances of all her- baceous species declined greatly after Baccharis formed a closed canopy at 2-3 yr, and little seed of herbaceous spe- cies was either dispersed into shrub stands or stored in the soil. Exclosures suggested that herbivory by small mam- mals in the closed shrub stands may be important in redu- cing the abundance of herbaceous species following shrub invasion of grassland. Key words: Grassland - Invasion - Herbivory - Seed pro- duction - Baccharis Community boundaries between grassland and shrubland in California are often remarkably stable for long periods (Davis and Mooney 1985). In a few cases, however, shrubs have been noted to invade adjacent grassland areas. In par- ticular, considerable range extensions of Baccharis pilularis ssp consanguinea (DC) C.B. Wolf have been reported by McBride and Heady (1968) and McBride (1974). Invasions appear to be episodic in nature, and the change from grass- land to shrubland dominated by Baccharis happens rapidly. Here we discuss the development of stands of Baccharis in an area where a recent range extension had occurred. At our study site, Baccharis has spread into neighbour- ing grassland extensively over the past 5-6 yr. The range of stand ages present allows detailed observation of changes in shrub biomass and structure and the effects of shrub invasion on the grassland vegetation. We wanted to assess how rapidly grassland species disappeared from the areas invaded by shrubs, how much seed of grassland species remained in the soil and how much transfer of seed from the grassland to shrub areas occurred. In addition we inves- tigated the role of small mammals in the vegetation changes observed. * Present address and offprint requests to: CSIRO, Division of Wildlife & Rangelands Research, LMB 4, P.O. Midland, W.A. 6056, Australia Methods Study site The area studied was located in the Jasper Ridge Biological Preserve, on the eastern slopes of the Santa Cruz Mountains in San Mateo County, Northern California. The study site was at an elevation of approx. 180 m on soils derived from greenstone (Franciscan formation). The area was grazed by cattle until 1960 but since then has been grazed only by deer and small mammals. Photographic evidence indi- cates that Baccharis has been present at the site for about 40 yr, but has been restricted to moister habitats such as gully bottoms and draws (Williams et al. unpublished work). However, in the 5-6 yr before our study commenced it had spread rapidly into surrounding grassland areas which have been described by McNaughton (1968). The site chosen for study consisted of a stand of old Baccharis surrounded by a series of younger stands which had estab- lished over the 6 yrs prior to the study. Shrub development We initially classified shrubs subjectively into age categories based on height and form. Five age categories were distin- guished at first, although the oldest catgory was later subdi- vided into 2 distinct categories. In stands of each age cate- gory we determined shrub density and estimated the amount of light reaching the ground surface by comparing measurements with a photon flux meter taken above the canopy and at ground level on an overcast day. We mea- sured ten shrubs in each category to give estimates of height, crown diameter and basal stem diameter. These shrubs were then cut down in November 1982, leaving only a few centi- meters of main stem above ground. The main stem of the oldest shrubs was not cut, but all branches arising from it were removed. We estimated the ages of shrubs by taking ring counts on the main stem or largest branch. Basal stem diameter was also measured. While age estimates can be considered accurate for most of the shrubs studied, ring counts on the main bole of old individuals were impossible, and age was estimated from the largest branch, giving a minimum age. We subsequently sorted harvested material from each shrub into leaves and stems. The entire shrub was analysed in all cases except for age classes 5 and 6, where we took a subsample of one main branch. Samples were then dried at 80 ~ C for 48 h and weighed.