Fire is a dominant force shaping patterns of plant diversity in Mediterranean-type ecosystems. In these biodiversity hotspots, including California’s endangered coastal scrub, many species remain hidden belowground as seeds and bulbs, only to emerge and flower when sufficient rainfall occurs after wildfire. The unique adaptations possessed by these species enable survival during prolonged periods of unfavorable conditions, but their continued persistence could be threatened by nonnative plant invasion and environmental change. Furthermore, their fleeting presence aboveground makes evaluating these threats in situ a challenge. For example, nitrogen (N) deposition resulting from air pollution is a well-recognized threat to plant diversity worldwide, but impacts on fire-following species are not well understood. We experimentally evaluated the impact of N deposition on post-fire vegetation cover and richness for three years in stands of coastal sage scrub that had recently burned in a large wildfire in southern California. We installed plots receiving four levels of N addition that corresponded to the range of N deposition rates in the region. We assessed the impact of pre-fire invasion status on vegetation dynamics by including plots in areas that had previously been invaded by nonnative grasses, as well as adjacent uninvaded areas. We found that N addition reduced native forb cover in the second year post-fire while increasing the abundance of nonnative forbs. As is typical in fire-prone ecosystems, species richness declined over the three years of the study. However, N addition hastened this process, and native forb richness was severely reduced under high N availability, especially in previously invaded shrublands. An indicator species analysis also revealed that six functionally and taxonomically diverse forb species were especially sensitive to N addition. Our results highlight a new potential mechanism for the depletion of native species through the suppression of ephemeral post-fire biodiversity.

Materials and methods

Study site

We conducted field experiments at a site in the Santa Monica Mountains National Recreation Area (34.151125, -118.964891) in stands of coastal sage scrub shrublands which burned in the May 2013 Springs Fire. The site last burned in 1993. The Santa Monica Mountains are adjacent to the highly urbanized Los Angeles metropolitan area and experience a steep gradient of anthropogenic N deposition. The study site where field plots were installed is located in the western end of the mountain range which receives moderate levels of exogenous N inputs, approximately 8 kg N ha^-1 yr^-1. Using aerial imagery and initial site surveys, we selected study areas of coastal sage scrub that were previously uninvaded by nonnative annual grasses before the fire, as well as adjacent areas at the ecotone of native shrublands and annual grasslands where both native shrub species and nonnative grasses were present before the fire (hereafter referred to as uninvaded and invaded, respectively).

Nitrogen fertilization

We installed plots (5m x 5m) in September 2013 (Fig. 1A) and monitored vegetation dynamics for three growing seasons. In a randomized block design, we treated plots  with one of four levels of N addition (0, 5, 15, and 30 kg N ha^-1 yr^-1) annually in the fall before the first rains of the growing season to simulate the accumulation of dry deposition to the soil surface throughout the dry season. Plots within each block were separated by a buffer of at least two meters. Plots were fertilized by hand-broadcasting a mix of granular calcium nitrate (CaN₂O₆) and urea (CO(NH₂)₂), an approach previously used to examine impacts of N addition in this ecosystem. We installed blocks that included replicate plots of each N treatment (n = 10) in stands of coastal sage scrub that were previously invaded by nonnative annual grasses (prior to the fire), as well as in directly adjacent areas of shrublands that were not. Invasion status (invaded or uninvaded) was determined based on aerial imagery and site surveys (both before and after the fire). Blocks in invaded and uninvaded shrublands were matched in terms of aspect and slope.

Vegetation monitoring

Each spring at the peak of the growing season (late April to early May), we sampled vegetation within all plots. We placed four 1m² subplots within each plot and visually estimated percent cover in a gridded plot frame of all plant species present. From these data, we calculated plant cover by native and nonnative plant functional groups, including for native shrubs, native forbs, native grasses, nonnative annual grasses, and nonnative forbs. We also surveyed entire plots for plant species richness to ensure we captured species not present within the smaller subplots sampled for plant cover.