Wildfires and climate change are having transformative effects on vegetation composition and structure, and post-fire management may have long-lasting impacts on ecosystem reorganization. Post-fire aerial seeding treatments are commonly used to reduce runoff and soil erosion, but little is known about how seeding treatments affect native vegetation recovery over long periods of time, particularly in type-converted forests which have been dramatically transformed by the effects of repeated, high-severity fire. In this study, we analyze and report on a rare long-term (23-year) dataset that documents vegetation dynamics following a 1996 post-fire aerial seed treatment and subsequent 2011 high-severity reburn in a dry conifer forest of northern New Mexico in the southwestern United States. Repeated surveys between 1997 – 2019 of 49 permanent transects were used to test for differences in vegetation cover, richness, and diversity between seeded and unseeded areas, and to characterize the development of seeded and unseeded vegetation communities through time and across gradients of burn severity, elevation, and soil-available water capacity. Post-fire seeding led to a clear and sustained divergence in herbaceous community composition. Seeded plots had much higher cover of non-native graminoids, primarily Bromus inermis, a likely contaminant in the seed mix. High-severity reburning in all plots in 2011 reduced native graminoid cover by half at seeded plots compared to both pre-fire levels and to plots that were unseeded following the initial 1996 fire. In addition, increased fire severity was associated with increased non-native graminoid cover and reduced native graminoid cover, native species richness, and species diversity.   This study documents a fire-driven ecosystem transformation from a former conifer forest into a shrub-grass system, reinforced by aerial seeding of grasses and high-severity reburning. This unique long-term dataset illustrates that post-fire seeding carries significant risk of unwanted non-native species invasions that persist through subsequent fires – indicating that alternative post-fire management actions merit consideration to better support native ecosystem resilience in the face of emergent climate change and increasing disturbance. Lastly, this study highlights the importance of long-term monitoring of post-fire vegetation dynamics, as short-term assessments will miss key elements of the full complexity of ecosystem responses to fire and post-fire management actions.

The vegetation data collection method followed the line-intercept method (Mueller-Dombois and Ellenberg 1974). Vegetation transects were approximately 50 meters in length and the data collected included herbaceous foliar and basal cover of all live and dead vegetation to the nearest cm. Plants were identified to the species-level, when possible, otherwise plants were grouped by their genus, or unknown plants were grouped by their growth form (see below). This line intercept method records the number of canopy-cover centimeters per species, and therefore allows for overlapping total vegetative canopy cover more than the total transect length. In cases where combined canopy cover for an individual species exceeded 100%, these species values were constrained to 100%. Canopy cover is reported as a proportion of the number of centimeters recording live foliar cover of a given species or growth form relative to the length of the transect. Bare ground is reported as a proportion of exposed soil, measured as the basal gap area between plant bases or other objects embedded in the surface (e.g., logs, rocks, litter) relative to the length of the transect.