BUSH LUPINE MORTALITY, ALTERED RESOURCE AVAILABILITY, AND ALTERNATIVE VEGETATION STATES

Nitrogen‐fixing plants, by altering the availability of soil N, potentially facilitate plant invasion. Here we describe how herbivore‐driven mortality of a native N‐fixing shrub, bush lupine (Lupinus arboreus), increases soil N and light availability, which promotes invasion by introduced grasses to the detriment of a native plant community. Soils under live and dead lupine stands contained large amounts of total N, averaging 3.14 mg N/g dry mass of soil (398 g/m²) and 3.45 mg N/g dry mass of soil (438 g/m²), respectively, over four years. In contrast, similar lupine‐free soil was low in N and averaged only 1.66 mg N/g dry mass of soil (211 g/m²) over three years. The addition of N fertilizer to lupine‐free soil produced an 81% increase in aboveground plant biomass compared to plots unamended with N. Mean rates of net N mineralization were higher under live lupine and where mass die‐off of lupine had occurred compared to soils free of bush lupine. At all sites, only 2.5–4.2% of the total soil N pool was mineralized annually. Soil enriched by lupine is not available to colonists while lupines are alive. The dense canopy of lupine shades soil under shrubs, reducing average photon‐flux density in late spring from 1725 μmol·m⁻²·s⁻¹ (full sunlight) to 13 μmol·m⁻²·s⁻¹ (underneath shrubs). Stand die‐off due to insect herbivory exposed this bare, enriched soil. In January, when annual plants are establishing, average photon‐flux density under dead lupines killed by insect herbivores was 370 μmol·m⁻²·s⁻¹, compared to the photon‐flux density under live lupines of the same age, which averaged 83 μmol·m⁻²·s⁻¹. The availability of bare, N‐rich patches of soil enabled nonnative annuals (primarily Lolium multiflorum and Bromus diandrus) to colonize sites, grow rapidly, and dominate the plant assemblage until lupines reestablished after several years. The N content of these grasses was significantly greater than the N content of the mostly native plants that occupied adjacent coastal prairie devoid of bush lupine. Between 57 and 70% of the net amount of N mineralized annually was taken up by introduced grasses and subsequently returned to the soil upon the death of these annuals. Even in the absence of further N inputs, we estimate that it would take at least 25 yr to reduce the soil N pool by 50%, indicating that the reestablishment of the native prairie flora is likely to be long term.