In Bromus mollis, a widespread annual grass in the summer-dry California grasslands, leaf senescence occurs during inflorescence development. Under field conditions in April, as soil was beginning to dry, midday conductances and 14C photosynthesis were highest in leaves near the apex of the plant, but lower older leaves maintained more positive leaf water potential ({leaf). Chlorophyll a, a/b ratio and carotenoid content were also greater in higher leaves. Over half of the 14C labeled in midmorning accumulated in the inflorescences by evening. A controlled experiment showed that soil drought results in midday stomatal closure and lower 4leaf, lower nitrogen content and less effect of leaf position on these parameters. In B. mollis, regulation of nitrogen and chlorophyll content in relation to leaf position may be ways of increasing carbon assimilation once self-shading has begun and soil moisture deficits are imminent. INTRODUCTION Annual grasses from the Mediterranean Basin invaded California's dry interior grasslands and replaced the native perennial bunchgrasses during the last century in response to overgrazing and human impact. Present-day dominance of these annuals has been attributed to their patterns of phenology and growth: vegetative growth in autumn and early spring is rapid and plants have high aboveground allocation. Then, there is a sudden and complete switch from vegetative to reproductive growth. This is accompanied by senescence of basal leaves, and mortality occurs just as summer drought begins (Gulmon, 1979; Ewing and Menke, 1983a, 1983b; Jackson and Roy, 1986). This developmental pattern is characteristic of plants classified as drought-escapers (Levitt, 1972). We examined physiological adaptations associated with the changeover from vegetative to reproductive growth in the annual grass, Bromus mollis L. Changes in carbon and nitrogen partitioning were expected to occur along with the development of inflorescences and the overtopping of basal leaves during the reproductive growth phase (Salter and Goode, 1967; Mooney et al., 1981; Mooney and Gulmon, 1982; Kirkham and Kanemasu, 1983). Our objective was to determine: (1) the amount of carbon moving into inflorescences compared to other structures; (2) carbon assimilation and water relations in upper and lower leaves, and (3) nitrogen concentration and chlorophyll content in foliage at different levels on the stem. Another aspect of this study examined carbon gain and water relations of B. mollis in response to low soil moisture during the seedfilling stage. METHODS Bromus mollis was studied during the seed-filling stage under two sets of conditions. Field measurements were made in annual grassland near Livermore, Calif., on 24 April 1984. Also, plants grown in large wooden boxes (2 x 4 x lm depth) under a clear plastic shelter to eliminate precipitation were studied on 14 May and 1 June 1984 in Berkeley, Calif. The annual grassland site was a dense stand of annual grasses dominated by Bromus mollis, Avena barbata and Lolium multiflorum. On the day of sampling (24 April), few green 1Present address: Department of Plant and Soil Biology, University of California, Berkeley, 94720. 2Present address: Botanisches Institut, Universitat Innsbruck, A6020 Innsbruck, Austria.