Shrub communities are valuable ecosystems that are protected under the Spanish Natura 2000 network, being the conservation and maintenance of these ecosystems strongly linked to a particular degree of management (Muñoz et al., 2012b, 2014). In Galicia (NW Spain), fuel reduction treatments are used in those communities to reduce the risk of high-severity wildfire and to meet other ecological and socioeconomic objectives (Vega et al., 2001). In the particular case of heathland communities, they are considered to be endangered by both atmospheric nutrient deposition and natural succession and that a high-intensity management is necessary for their conservation. Although research on the effects of fuel treatments on shubland communities has increased in recent years (e.g. Calvo et al., 2005; Fernández & Vega, 2014, 2016; Fernández et al., 2015), still remains the uncertaincy on how a wildfire combined with fuel treatment could affect the regeneration of an obligate seeder species.

In the present study, we investigated how a plant community dominated by Erica umbellata Loefl. (L.) responded to a backfire that occurred four years after application of different types of fuel-reduction treatments. The aim of the study was to detect any differences in shrub recovery in relation to fuel treatment (i.e. differences between treated and untreated areas or between types of treatment).

The treatment factor had no effect on total vegetation cover (Table 1). Before the wildfire, four years after fuel reduction treatments, the mean vegetation cover in the area treated by prescribed fire was 108.5%, compared with 107.6% in the areas treated by clearing and 105.9% in the areas treated by mastication, respectively (Fig. 1). Four years after the wildfire, the values ranged from 152% in the areas treated by prescribed burning or clearing to 129% in the area treated by mastication. The total vegetation cover was, on average, 153% in the control areas.

Separate analysis of each shrub species showed that the response differed depending on the species (Table 1). Treatment had no effect on the recovery of P. tridentatum or U. gallii (Table 1; Fig. 2). The treatment factor was significant for the E. umbellata and H. lasianthum cover (Table 1). For E. umbellata cover, clearing and mastication differed from the control (p < 0.0001 in both cases). For H. lasianthum cover, the three fuel treatments differed from the control (p < 0.05). Treatment affected grass cover, mainly resprouting of P. longifolium and A. curtissii (Table 1), with the responses differing significantly between all treated areas and the control (p < 0.01). However, there were no differences in grass cover in the three fuel treatments. No treatment effect was observed for forbs cover (Table 1).

In the treated areas, the backfire did not affect the recovery of the resprouting species (i.e. P. tridentatum and U. gallii), as also observed after an experimental fire in this area (Fernández & Vega, 2014) and supports previous research that demonstrated the high resprouting capacity of those species after perturbations (Reyes et al., 2009; Muñoz et al., 2012b) along with a possible increase in heat-stimulated seed germination (Rivas et al., 2006).

The lower E. umbellata cover in the areas treated by clearing and mechanical shredding compared with prescribed burning and control may be related to higher soil burn severity, as dead woody material may accumulate after mechanical treatments, but not after prescribed burning treatment (Fernández et al., 2013a). Fuel reduction may have favoured a higher heat penetration in the soil, thus promoting a reduction in the soil seed bank (Schimmel & Granstrom, 1996). Nonetheless, E. umbellata cover recovery was faster than observed by Fernández & Vega (2014) in a previous study after fuel treatments and a low severity experimental fire (4 %, on average 3 years after the experimental fire). The rapid recovery also contrasts with previously observed low rates of germination of this species after fire (Reyes & Casal, 2008), probably of lower severity of that of this study.

The greater presence of H. lasianthum, a minor species in the community, in the treated than in control plots after the wildfire may be due to stimulation of germination by heat (Fernández et al., 2013b). The observed increase in grass cover is consistent with previously observed increases in the abundance of herbaceous graminoids and non-graminoids with the frequency of perturbation (Vilá-Cabrera et al., 2008).

As observed in other heathlands in the region, some kind of management is necessary to preserve those ecosystems (Muñoz et al., 2012 a and b, 2014). In the community under study, dominated by an obligate-seeder species, it seems that a severe fire that substantially reduces the soil organic layer can be necessary for a successful regeneration. A similar response was documented in some South-African fynbos species (van Wilgen, 2013). Consequently, mechanical treatments seemed to be less appropriated for the management of these communities. This result could be extended to other similar systems although more research is necessary to confirm this.

Fuel treatments, particularly prescribed burning, were favourable for the E. umbellata regeneration, in the community under study that was senescent at the beginning of the study. Similarly, prescribed fire is used for Calluna vulgaris heathlands preservation (Davies et al., 2008).

From a management point of view, the fuel treatments considered did not have different effects on the recovery of heathland burned by a backfire. Shrub cover recovered quickly after fire in both in the treated and untreated areas.

Prescribed burning seems to be the most favourable treatment for enhancing regeneration of E. umbellata although other possible consequences of this treatment must also be considered (e.g. increased erosion risk and nutrient losses).