A low-intensity summer wildfire swept through some of Europe’s most treasured dry grasslands in eastern Hungary, and the vegetation barely flinched. Three years later, the burnt patches of pristine and restored loess grassland had largely reassembled themselves, with specialist species intact and no lasting takeover by weeds. But where firefighters carved emergency firebreaks into the earth, the story was starkly different: the deep soil disturbance left behind ruderal vegetation that still bore little resemblance to the surrounding grassland even three years on. The findings, published in Fire Ecology, expose a sharp trade-off between short-term fire safety and long-term biodiversity conservation in Central European grasslands.

The study took place in the Egyek-Pusztakócs Puszta within Hortobágy National Park, a flat former floodplain of the River Tisza that now supports a mosaic of loess grasslands, salt steppes, meadows and wetlands. Pristine loess grasslands on chernozem soils are among the most species-rich grassland types in the region and are listed as a priority habitat in Annex I of the EU Habitats Directive, yet they have been largely converted to croplands because of their fertile soils. Nearby, 93 hectares of loess grassland had been restored starting in 2006 by sowing low-diversity grass seed mixtures on former croplands. By the first year of the survey, these restored sites were 16 years old, hosted several target grassland species and carried weed cover below 5 percent.

Extensive wildfires broke out between 3 and 20 July 2022 after a prolonged hot and dry period, burning 227.8 hectares of pristine and 51.8 hectares of restored loess grassland. The fire was judged to be of low intensity based on field observations, the grazed condition of the vegetation and satellite imagery showing relatively low fire-induced spectral change, although intensity was not quantified directly in the field. Rapid fire spread posed a serious safety risk to nearby wetlands with high flammability, farms and residential areas. To halt the flames, emergency firebreaks were created by disk-harrowing, disturbing the soil to a depth of 15 to 20 centimeters and producing strips 8 meters wide.

The researchers set up a field experiment comparing five treatments: unburnt reference grassland, burnt grassland, and firebreaks under three restoration regimes. In each of three sites in both habitat types, firebreak sections were divided into 15-meter units. The first was left untreated after disk-harrowing, representing passive restoration. In the second, the soil surface was leveled and compacted by rolling to assist natural regeneration. In the third, after rolling, seeds of the competitive perennial grass Festuca rupicola were sown by hand at a density of 20 kilograms per hectare, using seed of regional provenance harvested from local loess grasslands. Rolling and sowing were applied in October 2022. In total, 180 permanently marked 1-by-1-meter plots were surveyed annually from 2023 to 2025, yielding 480 resurveyed plots, with all vascular plant species and their percentage cover recorded by visual estimation each May.

The team tracked total species richness, cumulative vegetation cover, and the richness and cover of species grouped by habitat affinity into grassland specialists, disturbance-tolerant species and weeds, as well as by life form and life span. Satellite data from Sentinel-2 imagery were used to compute the differenced normalized burn ratio, a spectral index for burned-area assessment, comparing pre-fire images from 3 July 2022 with post-fire images from 19 August 2022 and from the third growing season in July 2025. The immediate post-fire maps showed predominantly low spectral change, with values below 0.3 in both habitat types, supporting the low-severity assessment. By the third post-fire growing season, burn ratios in burnt and unburnt areas had become similar, indicating the earlier contrast had largely vanished.

In the burnt grasslands themselves, the fire’s effects proved moderate and temporary. Total species richness remained similar to reference levels throughout the study in both habitat types. Several grassland specialist species were significantly associated with burnt areas during the first two post-fire years, but these patterns disappeared by the third year and did not affect overall specialist richness or cover. Fire did not promote disturbance-tolerant or weed species in the long run; in pristine burnt grasslands, the richness and cover of both groups actually decreased over time. The only clear increasing trend was in short-lived graminoids in the restored sites. The authors interpret this as evidence that fire represents an endogenous disturbance to which these communities are adapted, triggering a short-term internal reorganization rather than permanent degradation.

The firebreaks told a different story. Although some grassland specialists established on the disturbed strips within the first year and their richness increased over time, the vegetation remained dominated by disturbance-tolerant and weed species throughout the three study years. Diagnostic weed species appeared in high numbers already in the first year, with 18 species across pristine and restored sites, and by the third year 10 weed species were still associated with disk-harrowed firebreaks. Specialist species richness did reach levels comparable to reference grasslands by year three, but their cover remained substantially lower, showing that the species pool recovers faster than vegetation structure. The researchers attribute the slow structural recovery to the fact that disk-harrowing completely destroys plant individuals within the disturbed area, while grassland specialists typically have sparse, transient seed banks.

Restoration treatment mattered. In sown firebreaks, Festuca rupicola established successfully in both habitat types, and by the third year perennial grass cover reached values seen in reference grasslands. The closing grass matrix effectively suppressed weakly competitive weeds, and the proportion of weeds declined substantially over time, although weed richness and cover in the third year still exceeded reference levels, suggesting complete suppression needs more time. Yet the dense grass cover came at a cost: reduced availability of establishment gaps and increased competition may hinder the recruitment of other grassland specialists, and species composition of sown firebreaks differed markedly from that of disk-harrowed and rolled ones. In restored sites, where specialists arrived from incoming seed rain, recovery was faster than in pristine sites, where specialist establishment initially benefited from a more developed seed bank but later stalled.

The contrast between the two disturbance types is central to the study’s conclusions. Natural disturbances such as grazing produce a network of small patches with shallow soil disturbance, whereas disk-harrowing affects deeper soil layers over large linear extents and causes mass mortality of plants through cutting and burial. The uneven soil surface left by harrowing created heterogeneous microsites that boosted total species richness, which was highest in disk-harrowed firebreaks, but simultaneously hindered recovery of natural vegetation structure and maintained weed populations. In pristine sites, rolling mitigated this effect, likely because leveling reduced microsite heterogeneity. The high abundance of unwanted species on firebreaks may also pose a risk to adjacent intact grasslands by acting as sources of weed infestation.

The authors argue that firebreak establishment should be applied cautiously and accompanied by appropriate monitoring and, where necessary, targeted management such as eliminating weeds before seed ripening or removing biomass to facilitate specialist establishment. They suggest that protection of farmland and residential infrastructure should instead rely on low-fuel areas such as productive fuelbreaks, where targeted grazing or mowing can reduce fuel loads in a nature-friendly way, potentially aided by emerging technologies like virtual fencing. When emergency firebreaks are unavoidable, the choice of restoration strategy should follow local conditions: passive restoration or assisted natural regeneration where target species are well supplied with seeds and weed pressure is low, and sowing the locally dominant grass where propagule availability is poor and weed risk is high. An important next step, the researchers note, is evaluating how these vegetation trajectories influence future wildfire risk by integrating ecological indicators with measurements of fuel load, fuel continuity and vegetation moisture.

Subject of Research: Vegetation recovery in burnt Central European dry grasslands and firebreaks under passive, assisted and active restoration treatments

Article Title: Trade-offs between fire suppression and biodiversity: vegetation recovery in burnt grasslands and firebreaks under different restoration treatments

Article References: Deák, B., Süveges, K., Gőri, S., Kapocsi, I., Gál, L., & Valkó, O. (2026). Trade-offs between fire suppression and biodiversity: vegetation recovery in burnt grasslands and firebreaks under different restoration treatments. Fire Ecology, 22(1), Article 128. https://doi.org/10.1186/s42408-026-00578-1

Image Credits: AI Generated

DOI: 10.1186/s42408-026-00578-1

Keywords: wildfire, firebreaks, grassland restoration, disk-harrowing, biodiversity, seed sowing, Festuca rupicola, Hortobágy National Park, loess grasslands, disturbance ecology, weed suppression, fire suppression