Disturbance regimes, like low-intensity fire, canopy gaps, and ungulate browsing, play a critical role in determining ecological composition and structure in temperate forests around the world.

Each disturbance (or lack thereof) can lead to unique plant communities, but we do not understand how combined disturbances change plant diversity and the resulting soil seed bank. Changes in the soil seed bank, which depend on the plants that survive post-disturbance, can then influence future biodiversity and succession.

We used a long-term experiment in West Virginia, U.S.A. that factorially manipulated low-intensity fire, deer exclusion, and canopy gaps. Thirteen years after disturbance initiation, we sampled the seed bank from each disturbance treatment.

We found that low-intensity fire led to increased seed bank density, with additional canopy gaps and deer exclusion each creating unique seed bank communities. Combined fire, canopy gaps, and deer presence led to high seed bank diversity and the most unique seed communities, while canopy gaps and deer had no effect on seed banks unless the area was previously burned. In contrast, combined fire, canopy gaps, and deer exclusion led to the lowest seed bank diversity of all treatments, reflecting the continued legacy of extant plants that grew immediately after disturbance. Seed communities were also distinct from extant understory species over thirteen years, regardless of disturbance treatment.

Each reintroduced disturbance combination left a unique legacy in the seed bank that will likely influence future forest reorganization following disturbances, adding to our understanding of how multiple disturbances influence forest succession and organization.

Synthesis: Forest disturbance regimes have changed around the world and are being restored or manipulated to support biodiversity. Reintroduction of disturbance leads to unique plant communities, but we do not understand how combined disturbances change the soil seed bank. Using an experiment that manipulates low-intensity fire, canopy gaps, and deer exclusion, we find that combinations of these experimental treatments leads to substantially different seed communities. These disturbance-altered seed banks will likely influence future biodiversity and successional patterns, highlighting how the restoration of disturbance can strongly and indirectly influence temperate forest community dynamics.

Site: We manipulated prescribed fire, deer presence, and canopy gap creation in four replicate Appalachian hardwood stands in central West Virginia, USA. We established this experiment in 2000 in the Western Allegheny Mountain ecological subsection using two stands in the Monongahela National Forest (39º06’ N, 79º43’ W) and two stands in the Fernow Experimental Forest (39º01’ N, 79º42’ W). Each stand was 60 to 90 years old and between 670 to 800 m in elevation. 

Experimental Design: Our experimental design was a split-plot factorial (Fig. 1), with each stand split in half and randomly assigned a burn treatment (burned or unburned). In each burned and unburned half stand, we established treatment plots (20 x 20 m, 400 m²) representing 8 disturbance treatments: 1) no burn + no deer + no canopy gap, 2) no burn + no deer + canopy gap, 3) no burn + deer + no canopy gap, 4) no burn + deer + canopy gap, 5) burn + no deer + no gap, 6) burn + no deer + canopy gap, 7) burn + deer + no canopy gap, 8) burn + deer + canopy gap. There were 8 replicates per treatment. Treatment plots were 20 m from one another, stand edges, and burn lines to avoid nonindependence and edge effects. One fire and canopy gap treatment plot could not be found in 2013, thus n = 63. We placed five permanent 1-m² sampling quadrats within each treatment plot. We collected seed bank samples at each corner of the five quadrats using a 5-cm long section of a 10-cm PVC pipe (Soil Volume: 392.5 cm³ x 4 = 1570 cm³ soil sampled per quadrat). All 20 soil cores per treatment plot were then pooled, mixed, and subsampled for use in emergence trials. Three subsamples were taken from each of the 63 treatment plot’s pooled soils and placed in separate 625 cm² square trays in a greenhouse (625 cm² x 3 = 1875 cm² soil per plot; 1563 cm³ x 3 = 4689 cm³ soil per plot), with 2.5 cm of subsampled soil placed on top of 2 cm of sterile sand in each tray. We watered all 189 trays (63 treatment plots x 3 subsamples) daily and occasionally rotated the trays to minimize any greenhouse positional effects (e.g., light, temperature). All germinants were identified to species or genera depending on life form, counted, and removed from the tray. After 5 months, we subjected trays to a 90-day, 5º C cold stratification period, after which they were returned to the greenhouse for another germination phase.

Analysis: For operational purposes we define seed species density as the total number of species found across the three trays representing a single treatment plot. Similarly, seed abundance is defined as the total number of germinants found across the three trays per plot. Classic Shannon diversity is calculated via the vegan R package (Oksanen et al. 2022).