Field surveys were carried out in three study sites in Sicily (Italy), all classified at a critical risk of desertification (Calvi et al., 2016; Figure 1; Supplementary Table S1). The first prickly pear orchard (PPR) is localized in Roccapalumba (Palermo province), whose landscape is dominated by clay soils and cultivated lands, especially annual crops, olive, and prickly pear orchards. The study site falls within the upper meso-Mediterranean upper dry bioclimatic belt (Bazan et al., 2015), at an altitude of 500 m a.s.l., on slopes less than 20%. Mean annual precipitation is 561 mm, mean annual temperature is 15.7°C, and soils are classified as Typic pelloxererts (Fierotti, 1988). The second prickly pear orchard (PPSP) is localized in Santo Pietro (Catania province), in hilly areas of south-eastern Sicily, dominated by sandy soils and covered by cultivated areas, including annual crops, vineyards, pasturelands, and hardwood forests and shrublands. The study site falls within the upper thermo-Mediterranean with a lower sub-humid bioclimate (Bazan et al., 2015), at an altitude of 262 m a.s.l., on slopes less than 10%. Mean annual precipitation is 690 mm, mean annual temperature is 17.0°C, and the soils are Typic xerochrepts (Fierotti, 1988). The third prickly pear orchard (PPC) is localized in Caltagirone (Catania province), in inner hilly areas dominated by clay soils and cultivated lands, subject to an upper thermo-Mediterranean lower sub-humid bioclimate (Bazan et al., 2015), at an altitude of 427 m a.s.l., on slopes less than 10%. Mean annual precipitation in this area is 690 mm, mean annual temperature is 17.0°C and the soils are Typic xerochrepts (Fierotti, 1988). All the surveyed prickly pear orchards were planted approximately 30 years ago. Most of the surface of the PPR and PPC orchards is currently managed, with the natural regeneration controlled through mowing; however, from 2005 onwards, local farmers have been leaving some rows unmanaged that were selected for sampling. PPSP orchard was totally abandoned after 2005, although it is currently accessed by horses, sheep, and cattle. The PPSP falls within the Site of Community Importance “Bosco di Santo Pietro” (ITA 070005), which hosts some of the most significant cork oak stands in Sicily.

Overall, 1,516 individuals from 25 different woody species were recorded in the sampling plots, with an average density of 53.7 individuals per 100 m⁻² (Table 1). The rarefaction analysis indicated that our sampling effort can be considered satisfactory to represent the plant community of recruits at our sites (Supplementary Figure S1). Colonization under prickly pear was widespread, with 94.6% of the plots containing at least one recruiting individual, while this percentage dropped to 61.6% under olives and to 22.3% in open areas. Average natural regeneration density under prickly pear was almost double that under olive trees (114 ± 99 × 100 m⁻² vs. 60.4 ± 76.4, respectively) and almost 30 times higher than in open areas (4.6 ± 9.3), with GLM indicating significant differences both considering all plant species and only shrubs and trees (Tables 2, 3; Figure 2). In turn, density and richness were higher under olives than in open areas. Prickly pear’s highest woody species density and richness were confirmed by the separated GLM models, indicating significantly higher values in all intra-site comparisons (Tables 2, 3; Figure 2).

Out of the 1,516 woody individuals, 202 (12.9%) were shrubs and trees higher than 100 cm, belonging to 15 species, and 94.4% of them occurred under the prickly pears. Indeed, only six plots in open sites and five plots under olives hosted such recruiting individuals, which were, conversely, widely spread over 117 prickly pear plots, accounting for 1.4, 10.9, and 19.7% of all woody individuals in the three habitats, respectively. Recruits were also significantly higher and larger under prickly pear than under olives and open areas (Table 4), including 53 individuals higher than 300 cm and with basal diameters up to 39 cm (Figure 3). Such differences were also confirmed by the separated GLM models (Table 4; Figure 3).

Wild asparagus (Asparagus acutifolius L.) was the most abundant species (28.9% of all the individuals) and, together with olive (Olea europaea L. s.l., 18.2%), mastic (Pistacia lentiscus L., 15.5%), and brambles (Rubus spp., 2.6%), occurred in all three habitats (Table 1). Seventy-six individuals of three oak species, i.e., downy oak (Quercus pubescens Willd. s.l.), cork oak (Q. suber L.), and holm oak (Q. ilex L.), were observed, but only under prickly pears. Downy oak was the most abundant oak species (80.3%), occurring at relatively high densities (19 × 100 m⁻²) and even reaching heights up to 7.5 m and diameters of 21 cm in PPC. Olives, almonds (Prunus dulcis (Mill.) D.A. Webb), mastic and downy oaks, (Quercus pubescens) accounted for 91% of the established individuals, although only the first two occurred in all prickly pear orchards. The large majority of the recruits were from fleshy-fruited species (92.6%), indicating an active role of animal seed dispersal networks.

Since we could not perform a statistical analysis of the influence of the nearest forest stands due to low sampling sites, we report here, and in Supplementary Table S2, a descriptive assessment. In PPC, the area with almost all oaks recruited, there were four downy oak stands with average size of 3.7 hectares, occurring 500 m away, and no wildfires were registered in these areas in the last 14 years. In PPR, there were three downy further than oak stands with average size of 8 hectares, but all of them were further than 1.5 km and had burned at least twice in the last 14 years. In PPSP, we found two Q. suber forests (average size of 10 hectares) and one large Q. ilex forest (>10 hectares), although 1.8 km distant.

Recruits were widely distributed under prickly pear in all three study sites, with 94.6% of the 150 sampled individuals hosting at least one woody species. Olive trees also showed a high frequency of woody recruits (62%), in comparison to open areas (22%). However, virtually all fully established (≥1.0 m high) shrubs and trees were found under prickly pear. In a study comparing the natural regeneration under a native shrub [Retama sphaerocarpa (L.) Boiss.] and paired open sites along an environmental gradient in Spain, just about 10% of the 1,263 Retama shrubs had one recruit, for a total of 211 woody individuals belonging to four species (Andivia et al., 2017). The authors also observed that Retama’s facilitative effect was negatively correlated with aridity and herbivory pressure. Due to the CAM metabolism, the high water-use efficiency, the shallow root system, and the high level of herbivory protection, prickly pear is expected to cause a positive correlation between increasing aridity and herbivory and its facilitative effect. Olive trees showed significantly fewer woody recruits than prickly pears, particularly those lacking established shrub and tree species (Figure 3; Table 4), possibly resulting from a higher competition for light and space because abandoned olive trees often have a high density of lower branches and root shoots. Conversely, prickly pear has a more open architecture, giving support to lianas and allowing the full growth of trees and high shrubs. A facilitative effect was also observed in the dynamics of biogroups in Italian mountain ecosystems, where junipers (such as Juniperus deltoides R.P. Adams and Juniperus hemisphaerica Presl) have been found to act as the development center for several woody species (Lapenna and Fascetti, 2010; Pedrotti, 2019). Indeed, a global meta-analysis indicated that shrubs tend to provide higher facilitative effects than trees, especially when considering later developmental stages (Gómez-Aparicio, 2009), although such ontogenetic trade-off was also observed within the same shrub species (Rolo et al., 2013). Indeed, recruitment was significantly higher under olives than in open areas, and the establishment of lianas such as Asparagus acutifolius and wild madder (Rubia peregrina L.) was widespread there. Interestingly, Andivia et al. (2017) also found that Asparagus acutifolius was the most abundant species occurring under Retama shrubs. The second most abundant recruiting species was olive, occurring in all prickly pear orchards, with some individuals higher than 6 m. A high olive seedling density has also been verified under Mediterranean pine plantations, although saplings and adults were very rare (Badalamenti et al., 2018). In turn, olive recruits were absent or rare under other olives and open areas, indicating both high intra-specific competition and susceptibility to drought and herbivory.

The positive role of prickly pear in the ecological restoration of semi-arid ecosystems has been previously acknowledged (Le Houérou, 1996; Neffar et al., 2013; Génin et al., 2017; Neffar et al., 2018) and the utility of prickly pear to improve soil conditions in order to favor the cultivation of other species has been highlighted in the past, but with exclusive reference to agricultural contexts (Nocito, 1844). Due to its peculiar plant structure, the prickly pear could also be a useful plant for combating erosion in riparian contexts, where, moreover, its invasive potential is effectively zero (Stavi, 2022). However, its possible facilitative role in the natural regeneration of Mediterranean woody species has been barely documented. In one of the few study cases, prickly pear has been reported to aid the development of the argan tree (Sideroxylon spinosum L.) in pre-Saharan Morocco, with an increasing effect of old prickly pear individuals resulting in soil organic matter accumulation (Génin et al., 2017; Hassan et al., 2019). Prickly pear has also been facilitating the natural regeneration of common walnut (Juglans regia L.) and almond (Prunus dulcis) in Mediterranean agroecosystems (Bueno et al., 2020a; Badalamenti et al., 2022). Recently, a study in the savanna ecosystem showed that the invasive Opuntia stricta can create fertility islands that facilitate the establishment of native plants, improving soil abiotic and biotic conditions (Novoa et al., 2021). In another study, Oduor et al. (2018) found that the abundance and diversity of native species were not affected by prickly pear invasion, regardless of the cover percentage, showing that the negative effects exerted by alien plants are strongly context-dependent.

The rarefaction analysis indicated that our results are representative of the plant community of recruits present at our study sites (Supplementary Figure S1). The vast majority of the woody species found in our surveys (Table 3) have fleshy fruits and rely on vertebrates, particularly mammals and especially birds, for seed dispersal (Bueno et al., 2021). This evidence implies a high incidence and effectiveness of animal-mediated seed dispersal networks. This is particularly the case for birds, as they tend to avoid open areas or to stay on the ground, relying both on natural and artificial perches (Pons and Pausas, 2006; La Mantia et al., 2019). Although we did not assess the seed rain, our personal observations clearly indicate that birds frequently used both olive trees and prickly pear individuals. Indeed, during our surveys, we observed many seed dispersers, such as warblers (Sylvia spp.), thrushes (Turdus spp.), robins (Erithacus rubecula), corvids (Corvus cornix, Pica pica, and Garrulus glandarius), pigeons (Columba palumbus), and starlings (Sturnus spp.) perching or foraging inside the prickly pear orchards. Due to its quick development, prickly pear can soon offer suitable perches for birds, consequently fostering secondary succession underneath. Since herbivory and trampling are key biotic factors affecting the recruitment of woody species, the physical protection offered by a nurse plant is of paramount importance (Padilla and Pugnaire, 2006; Andivia et al., 2017). The PPC and PPR orchards are protected to prevent the access of large animals, whereas the PPSP and respective olives and open areas are frequently accessed by livestock (cattle, horses, and sheep), while wild herbivores such as deer are not present. Since spineless prickly pears are the most used varieties in our study sites, the main protective mechanism is the physical barrier created by the cladodes, particularly when they become hard with aging. Despite such protection was also provided by olives and can be offered by other native shrubs (Smit et al., 2008), the multi-stemmed architecture of prickly pear, similar to a candle holder, seemed to represent a great trade-off between protection and space and light competition (Rolo et al., 2013). Effectively, we observed several recruits showing a typical tree habit with a streamlined trunk, in comparison with the shrubby aspect of recruits in open areas that is typical of sites under high herbivory pressure.

Oak forests (dominated by Quercus spp.) represent the late successional stages in some areas of the Mediterranean basin, including the surroundings of our study sites. Since oaks are highly susceptible to recruitment limitations and are often lacking in reforestation interventions, they are frequently used as indicators in facilitation studies (Acácio et al., 2007; Gómez-Aparicio, 2009; Bobiec et al., 2018). The main abiotic constraints hampering the natural regeneration of Mediterranean oaks are poor soil conditions and summer drought. These two conditions may not only kill the seedlings but also affect the viability of acorns when moisture content drops below 26% (Acácio et al., 2007; Ganatsas and Tsakaldimi, 2013; Matías et al., 2019). Light limitation is another relevant factor, with reduced development in high-tree-density plantations or dense shrublands (Pausas et al., 2006). To overcome these limiting factors, prickly pear was found to be of crucial importance. Indeed, not only did we find a high oak seedling density under prickly pears, but they also hosted all the established oak individuals that reached up to 7 m high, indicating a good balance between soil organic matter content, water availability, space, and light competition (Bautista-Cruz et al., 2018; Hassan et al., 2019). However, our results also suggest a central role of oak seed limitation conditioned by the characteristics of the nearest forest fragments (especially size and distance) because the vast majority of high oak individuals were found in PPC. Jays and the wood mouse, both observed in our study sites (Cairone et al., 2020; Bueno et al., unpublished data), are the main seed dispersers of Mediterranean oaks. They act at very different spatial scales, with wood mice operating at short distances (10–50 m) and jays at medium distances (approximately 300 m), with rare long-distance dispersal events (Gómez, 2003; La Mantia and Bueno, 2016). Despite patches of oak forests being present in the surroundings of all the study areas and relatively large oak forest patches (up to 18 ha) being present inside the 12.5 km² surveyed area, only at PPC were they within the mean seed dispersal range of the Eurasian jay (Garrulus glandarius) (as far as 500 m), which seemed to be one major factor driving the highest oak density there. Interestingly, some oak individuals in PPC were already reproductive (Figure 4), indicating a further successional step where seed dispersal from outside may become increasingly less important. Conversely, the very high fire frequency affecting oak forests in the surroundings of PPR is a major detrimental factor, as recurrent wildfires generally curb seed production for several years (Trabaud and Galtié, 1996; Badalamenti et al., 2020b), therefore reducing the chances of long-term seed dispersal. The suitability of prickly pear for oak seedling establishment, as widely observed in PPC, would suggest, hence, that the direct seeding of acorns could be successfully implemented in PPR and PPSP sites in order to overcome seed limitation.

The limited number of studies evaluating prickly pear as a restoration tool largely depends on its prevalent use as a productive crop, so unwanted plant species in prickly pear orchards are usually removed. For this reason, we did not find other suitable areas for our research, which limited the number of sampling sites. However, we clearly recognize that the major constraint that may limit the use of prickly pear as a restoration tool is its invasive potential in the Mediterranean, where it represents a serious ecological threat, particularly for chasmophytic vegetation in rocky habitats (Vilà et al., 2003; Guarino et al., 2021). Even if in our study areas and in nearby abandoned fields and forest patches (e.g., pine and eucalyptus reforestation, oak forests, and maquis), we found no evidence of invasive behavior, we do not recommend its use in areas where it is not already present or if careful management is not feasible. From this side, it should be kept in mind that prickly pear is an agricultural crop already covering millions of hectares and sustaining the economy of entire regions, so that in many areas, including Sicily, there are no restrictions on its cultivation. Most importantly, its management is really feasible, as knocking down all the first fruits when they are unripe is a common agricultural practice made by local farmers, preventing their diffusion through seed dispersal. Prickly pear is also highly sensitive to shading, so that the cover by adult tree individuals, once living underneath its canopy, naturally prevents its future development. In turn, when native woody plants are fully established, prickly pear can be cut, and the residuals can be used to feed livestock (another common practice), mulch, increase soil organic matter, or be used as a natural biogel for reforestation purposes (Le Houérou, 1996; Vimercati et al., 2020; Stavi, 2022). Considering the several biotic and abiotic constraints hampering the restoration success of Mediterranean degraded areas, particularly in arid and semi-arid climates, our results indicate that prickly pear may represent a valid tool to facilitate the restoration of Mediterranean woodlands, provided that its invasive potential is carefully considered.